For several years now I’ve had readers ask me from time to time about my opinion of the current media and investment frenzy surrounding so-called Artificial Intelligence (AI). The first thing I’ve had to explain in each of those cases is that the label is a misnomer; generative large language models, to give these bloated creations of the software industry their proper name, are certainly artificial but they’re not intelligent in any meaningful sense of the term.
Generative LLMs churn out statistically likely sequences of words, pixels, or computer code. That’s what they do, and it’s all they do. They differ in complexity but not in principle from those amusing programs that generate pirate talk or postmodernist academic papers on demand. Influential computer scientist Emily Bender, in an important 2021 paper, labeled generative LLMs “stochastic parrots,” and the label’s an accurate one; for those who don’t speak the jargon, “stochastic” means “producing a statistically random output.” (I hope I don’t have to explain to anyone what a parrot is.) There’s nobody home in these so-called AIs, just a program that churns out sequences that imitate the material on which they are trained.
That isn’t useless, by a long shot. Quite a bit can be accomplished by generating statistically likely but unpredictable sequences of words or pixels or computer code. Yet the gap between this and the expectations that have been loaded on the AI phenomenon is at least as significant as the actual potentials of the technology. Listen to the promoters and true believers of AI hype and you can count on hearing claims that LLMs will be conscious, self-aware beings in a matter of months, that they will soon be far more intelligent than human beings, and that once this happens AIs will usher in Utopia if they don’t decide to exterminate us all first.
In point of fact, all of this is hogwash. Outside of a narrow range of uses, generative LLMs perform very poorly. Loud predictions of vast economic payoffs to companies that adopted them have turned out to be so much hot air, and the best LLMs on the market still perform very poorly indeed when confronted with genuinely new challenges—the sort of thing that human intelligence takes easily in its stride. A recent study, one of many, thus showed that LLMs can handle less than 3% of business tasks in the real world.
The same issues have become notorious in the publishing field. One of my publishers noted the other day that his firm fields plenty of LLM-written manuscripts. They are instantly recognizable as such; they are also unpublishable garbage. (The publisher has a snappy rejection email for these saying, “We publish people, not robots.”) Yet money keeps on pouring into the AI phenomenon, yielding a vast buildout of the gargantuan, expensive, energy- and water-hungry data centers that are needed to deploy LLMs at scale.
It’s quite a spectacle, and has led many commenters to wonder if we’re in the middle of another delirious speculative bubble on the scale of the one that popped so disastrously in 1999. The parallels are striking: the internet boom that burst a quarter century ago was set in motion by a genuinely transformative shift in technology, but that reality was buried in vast amounts of hype and a cascade of malinvestment that paid no attention to economic reality. Inevitably, the bubble popped, thousands of businesses and millions of investors went broke, and the frantic attempts by governments to paper over the mess led to the real estate bubble and bust of 2004-2009.
That could still happen. Still, I’ve come to think that there may be something far more important going on in the current frenzy over AI than the sort of common or garden variety speculative boom that so reliably separates fools from their money. To understand what I think may be going on here, however, we’re going to need to talk about catabolic collapse.
Longtime readers of my blog posts know about that, or at least have heard of it. Those without that background will want to know that it’s the subject of a paper I wrote in 2004, “How Civilizations Fall,” which you can read online here. Behind the paper lay years of research into the way that civilizations unravel. Despite the usual assumptions, this isn’t a fast process. It unfolds over one to three centuries of repeated crisis periods separated by intervals of stabilization and partial recovery. The question I wanted to explore was why this happens.
The answer’s actually quite simple. Civilizations always build more stuff than they can afford to maintain. By “stuff” I mean anything that an economist would call capital: buildings, roads, trading systems, social networks, educational institutions, transport technologies, ideologies that encourage people to contribute to the system, you name it. If it’s produced by human labor and has to be maintained by inputs of labor and resources, it’s capital—or, as I like to call it, stuff. Different civilizations fixate on different kinds of stuff, but every civilization produces a lot of stuff: usually as much as its resource base will allow.
The difficulty, of course, is that once built, stuff has to be maintained, which uses up labor and resources. Meanwhile there’s always a demand for new stuff, which also uses up labor and resources. Sooner or later, depending on how big your civilization’s resource base is, the demands on your labor force and resource base rise high enough that you can’t keep doing both. That’s when crisis arrives.
I borrowed terms from biology to provide labels for these stages. In any living body, including yours, the process by which new tissues are produced is called anabolism. (Think of anabolic steroids, which build muscle mass.) The corresponding process by which old tissues are broken down is called catabolism. The crisis that comes when a society no longer has the available labor and resources to meet the need for new production, and still maintain all its existing stuff, is a catabolic crisis. As the term suggests, what happens in that crisis is that a lot of old stuff gets torn down or just abandoned, so that the labor and resources that once went into maintaining it can be freed up for other uses.
You might think that civilizations, like living bodies, would simply balance catabolism against anabolism and keep things from going into crisis. Some human societies do this, but the kind of complex urban societies we call civilizations aren’t good at it. In particular, they reliably use the old stuff as infrastructure and build new stuff on top of it, and then more new stuff on top of that. This is why, for example, New York City still relies on a water system that was already obsolete in 1950. It’s also why the United States still relies on New York City as a port, marketplace, and population center when it’s been falling apart for many decades, and costs far more to maintain than it returns in terms of value to our society.
Of course politics, ideology, and other human factors come into play and make things even more complex. The smart thing to do with New York City in the 1950s would have been to tear it down one borough at a time, haul away the rubble, replace the decrepit infrastructure, and then rebuild. That was actually discussed, but it never got beyond the talking stages for political and cultural reasons. Instead, New York City will continue to be a crumbling, dilapidated, absurdly expensive mess until it’s abandoned, stripped of metal and other valuable resources, and left to the slowly rising oceans a century or two from now.
This isn’t a new story. Rome was a city with vast public buildings and apartment blocks seven stories high in 300 AD; by 600 AD the Forum at the center of the city was a goat pasture and the apartment blocks were distant memories. Plenty of other great cities of the past went exactly the same way, for the same reasons: the civilizations that built them kept pouring labor and resources into maintaining them, until they couldn’t.
It’s possible for a civilization can go through a catabolic crisis without collapsing. A lot of capital gets converted to waste, some of it abandoned, some stripped of raw materials for other uses. The United States went through a catabolic crisis of that kind in the 1970s and 1980s, and you can still find the dilapidated remnants of old factories and the former rights-of-way of disused rail lines all over the country. It was a wrenching transformation and ruined a lot of lives, but the nation survived it. It takes certain other factors to turn a catabolic crisis into the opening round of catabolic collapse, the process by which civilizations fall.
That’s not what I want to discuss here, however. Here we’re talking about the ordinary sort of catabolic crisis, the kind in which it stops being possible to keep covering the maintenance costs of unproductive capital, and so a lot of that capital is allowed to become waste so that labor and resources can be redirected to more productive uses. Yes, I’m suggesting that one of these is imminent in the United States and throughout the Western industrial nations.
There are very good reasons to think this. The global economy that funneled vast amounts of unearned wealth to the United States and its inner circle of allies is breaking apart into separate economic blocs; the status of the US dollar as default reserve currency is waning, though in the usual way this is happening a little at a time, not all at once. Other factors even less fashionable to discuss these days are placing sharp burdens on the available supply of labor and resources. Those will wait for a later post; what matters now is that something has to give.
This is where my deliberately broad definition of “stuff”—if you prefer, capital—is crucial to keep in mind. Stuff doesn’t have to consist of buildings, or roads, or New York City. It can also be organizational capital. An army is capital in this sense; so is a school, whether or not it meets in a building; so is a religion. So, finally, is a bureaucracy.
Here in the modern industrial West, we have far and away the largest bureaucratic sector in the recorded history of the planet. Our governments, local, regional, and national, are crammed to the bursting point with bureaucrats. What’s more, so are the office towers of our corporations and the headquarters of our nonprofits and nongovernmental organizations. These are so close to identical that personnel constantly move from one to another without the least disruption, and all can be seen as part of a single gargantuan bureaucratic structure, the most distinctive creation of industrial society over the last three quarters of a century or so.
It’s a very expensive creation. Here it’s crucial to remember that “stuff” isn’t just material objects such as office buildings, computer networks, yellow sticky notes, and all the other impedimenta that play roles in a modern bureaucracy. The ample salaries and benefits paid to bureaucrats are another part of the cost. So is the entire system of modern higher education, which abandoned most of its previous commitments to become a factory for manufacturing bureaucrats. So are the regulatory burdens loaded by government bureaucrats on more productive economic activities, and the costs of the pervasive corruption and graft that runs all through all these systems, from casual pilfering of company property to the multimillion-dollar bribes that routinely change hands at the upper end of the pyramid.
What makes all these costs even more galling than they would otherwise be is that the bureaucratic systems contribute so little and place so many burdens on productive economic activity. There is certainly such a thing as too little bureaucracy, but the United States left that behind by 1936 or so and has never looked back. My late wife, a bookkeeper during most of her working life, used to come home day after day with stories of the absurd regulations and heavy financial burdens loaded by city, county, state, and federal governments on the small businesses whose books she handled. In a post here two years ago I talked about lenocracy, literally “rule by pimps,” in which every activity is loaded with fees to the breaking point by corporate and government middlemen who contribute nothing but take an ever-increasing cut of the proceeds. That’s the standard strategy of bureaucracies these days.
In addition, these days our bureaucracies are not merely expensive and burdensome, they’re also stunningly incompetent. There’s a reason why NASA has to hire private contractors to put people into orbit; despite more than ample budgets, the bureaucracy that once put bootprints on the Moon hasn’t been able to design and build a working spacecraft in more than a quarter century. This is typical. Corporate and nonprofit bureaucracies are no better; need I cite Hollywood, which has completely lost the ability to make movies that audiences can enjoy?
None of this is accidental. As privileged classes become comfortable and decadent, they always try to shield their members from having to deal with the consequences of their own actions. That’s why you hear so much talk these days about “failing up:” no matter how miserably bad a job you do, if you belong to the bureaucratic class, you don’t have to worry about your career. Even if the firm you run crashes into ruin as a result of your idiocy, you’ll inevitably find a well-paying post elsewhere. The downside, of course, is that people thus sheltered never learn from their mistakes, and the institutions that shelter them end up mired in hopeless incompetence.
Bring all these factors together with the potential for catabolic crisis and the implications are obvious. Of all the ways to cut maintenance costs in today’s society without doing any significant damage to productive economic activity, far and away the most effective is to gut government, corporate, and nonprofit bureaucracies, freeing up the billions upon billions of dollars a year they cost (and the corresponding inputs of labor and resources) so that these can be put to genuinely productive uses, while unburdening the economy of the countless petty rules and fees that today’s bureaucracies load on them.
The one thing that makes this otherwise straightforward project difficult is that power is very diffuse in today’s industrial nations, and much of it resides in exactly those same bureaucracies that most need to be pruned. It’s indicative that the relatively modest steps the Trump administration has taken to trim the size of the federal bureaucracy and stop the wholesale looting of federal funds by corrupt politicians and nonprofits have unleashed shrieking rage and bloodcurdling threats of vengeance from leading members of the bureaucratic class. Carrying out comparably modest decreases in the corporate and nonprofit worlds, not to mention the much more extensive layoffs and shutdowns that will likely be needed, would be politically very difficult, unless it can be done in a way that seems inevitable, impersonal, and praiseworthy.
This is where the AI phenomenon enters the picture.
Our society has made a surrogate religion out of faith in progress. No matter how unsatisfactory something is, if you can convince people that it’s the wave of the future, a surprising number of them will accept it without quibbling. If you want an example, compare the frankly lousy sound quality on a phone playing streaming music with the vivid clarity of sound you get from a fifty-year-old stereo playing vinyl records. The step from records to CDs involved one sharp loss of sound quality, the step from CDs to streaming audio on computers a much greater one, but only a few eccentrics refused to make those changes. Even now, as vinyl records are resurgent, a vast number of people still put up with flat, tinny sound from phone speakers when a very modest investment would give them the chance to enjoy music much more. Why? Because they’ve been conned by the myth of progress.
The hoopla around AI, I’ve come to think, is intended to exploit that same reaction. Business media is full of stories these days about the vast number of white-collar jobs that will be going away forever in the near future as LLMs hit their stride. There’s a certain amount of fussing, to be sure, but nothing like the violent rage and hatred the Trump administration has gotten by making modest cuts in the federal bureaucracy. Most of the people who are likely to be affected are staring at AI-related stories like deer at the headlights of an oncoming truck. My guess is that this is no accident.
Ironically, it doesn’t matter in the least that LLMs can’t actually accomplish most of the tasks that human employees do. An immense share of the work done by today’s bureaucrats doesn’t need to be done at all, and an even larger share can be done quite well by machines that can produce statistically likely sequences of words that imitate human thought. Such work as is actually needed can be quietly outsourced to cheaper human labor, and of course this is already happening; a sour joke in today’s computer scene has it that AI actually stands for “Additional Indians,” because several companies have been caught using boiler rooms in Mumbai and Kolkata to do work that was supposedly being handed over to LLMs.
If I’m right, over the months and years immediately ahead, we can expect to see any number of triumphalist pieces in the mass media announcing that this or that department of this or that government, corporate, or nonprofit bureaucracy has been replaced by a shiny new AI. Governments at various levels will doubtless make various pro forma gestures in the direction of retraining the surplused bureaucrats, but those of us who remember how laid-off factory workers were treated forty years ago know better than to believe a word of it. Meanwhile, an entire world of businesses that supported themselves by providing services to cubicle inmates, from Pilates coaches to baristas and more, will also be left scrambling for survival.
Those bureaucrats I know who have a clue are already getting themselves retrained for jobs that produce real goods and services; they’re getting ready to become electricians, farmers, small business owners, and the like. Too many others are doing nothing to get ready for the wrenching changes ahead. I would encourage all my readers to assess the work they do to support themselves, remembering that if that job can be done by a stochastic parrot, they may not have a job five years from now. I would also encourage them to see what skills and talents they have that might be suited to small businesses. Once the leaden weight of excess bureaucracy gets lifted, there may be a great deal more room for small businesses than there has been in living memory. Still, we’ll just have to wait and see how it all plays out.
Hello JMG,
Thank you for your post , it is an interesting insight that makes sense: I saw many mainstream articles years ago about how ai would replace a high percentage of jobs , before ai become the thing it is today. I was wondering why the elite was pushing for that .
I have these comments:
1*The ai specialist I am following says that LLMs are only one section of AI in Computer Science and its current usage is brute force . Yann Lecun for example has said that it should be configured in a different way so that it is far less energy intensive and smarter for the algorithms . Other aspects of Computer Science receive much less attention and are just valuable
2* I remember ten years ago when the big IT corporations’ executives came out and said the future was in the cloud instead of on-premise. Then we discussed it with other IT professionals and we agreed this would create vast security issues, unnecessary complexity , and data loss for companies. Nevertheless the decision was made for strategic, investment and political purposes and we got the good , the bad and the ugly of cloud technology. It seems the same is happening now with AI data centers on an even bigger scale – whether we like it or not, and businesses and workers will have to live with the consequences.
4* there has been a few vast data theft in european government administrations this year. I thought this showed a level of incompetency compared to the level of responsibility they have with our personal and health data.
5*Regarding inefficiencies in organizations like corporations and bureaucracies, I have heard people say that most of the work is done by 20 per cent of people, and I agree. The thing is that it would not work with just those people because it is like trying to be an athelete with a very lean body, only eats healthy things and train rigorously. How long can groups of humans realistically behave like athetes in their lives ? What other parts of their lives will be neglected too much ?
Even the most dedicated groups become ‘ out of shape after some years ‘ . That level of performance is not sustainable . My explanation is not perfect but my intuition is that removing all the inefficient people from society is not going to work .
An exemple of a society that is under permanent stress, overwork at 80 hours per week, and high performance is South Korea . Their source of stress may be North Korea, but do we want to live like that for decades ? I had an American friend who was married to a South Korean woman, and he said ‘ it was hell ‘ .
I would rather live with less stuff and not have to live like a stressed-out zombie my whole life .
Just a critique of the current push to have everyone maintain living standards at all costs.
Good article. I only have a thought having to do with “The step from records to CDs involved one sharp loss of sound quality”.
Not really true, there was no loss in the quality of the sound, though there certainly is an argument between people who miss the crackles in the old vinyl and those who don’t (I’m often in the first camp precisely because I think there is such a thing as too much perfection in the arts), but provided you played both platforms in a good hifi system CDs perform better than records.
The following is true though, “the step from CDs to streaming audio on computers [involved] a much greater [loss]”, but again, the reason for this it’s not on the streaming part of the equation but on using the computer for reproduction.
I used to agree with this second half of the paragraph until a friend of mine who works in a high-end hi fi shop played me some music through a system in the shop (equal in price to a large family car) and I noticed he was playing vinyl, CDs and directly from Spotify with no difference in quality.
Often though, people who’ve been risen on the music of the 50s to 90s era will find some modern music unlistenable for reasons of their sound but that has to do with signal compression practices at the mixing and mastering stages in the recording studio which prioritise reproduction through tiny speakers and then sound awful on a hi fi or good headphones. But that’s a conscious decision.
If anyone doubts what I’m saying they only have to hear musicians of the analogic era commenting on the advent of digital music, for instance David Gilmour talking to Rick Beato, “when I listened to Pink Floyd’s first mix with digital means I never looked back”.
Wow, this is pretty much off topic, sorry!
This is an interesting essay.
I’m curious though, why you feel the weight of bureaucracy will be lifted. Surely the artificial bureaucrats can churn out just as many regulations as the living ones. If anything, more as they have no need for sleep and other human constraints.
While I would very much like to see the catabolism minus crisis, it is hard for me to see a good outcome here.
California and Colorado just passed a law requiring all operating systems to require age verification. How the **** are they going to get old Windows XP / Windows 7 running in random hospitals or mainframes like z/OS running in banks to comply with their age verification law? Let alone the millions of random Linux and BSD distributions sitting in servers and the cloud and in embedded / IoT systems? Just another example of clueless government bureaucrats making things impossible for everybody. If the California and Colorado politicians who passed the law succeed at enforcing the law, most businesses will be forced to abandon their computers and go back to pen and paper.
It seems like replacing most of today’s bureaucrats with LLMs won’t even require massive builds of data centers powered by new nuclear plants/perpetual motion machines. The Internet will become less usable due to AI slop but most people will quit using the Internet unless they absolutely have to over the coming years so it won’t matter that much. However, a lot of printed media will be LLM-generated so it might be a good idea to avoid most nonfiction books written after 2021.
You had to pick this topic, of all the topics out there.
>The hoopla around cocaine, I’ve come to think, is intended to exploit that same reaction. Business media is full of stories these days about the vast number of white-collar jobs that will be going away forever in the near future as cocaine use hits its stride. There’s a certain amount of fussing, to be sure, but nothing like the violent rage and hatred the Trump administration has gotten
https://inv.nadeko.net/watch?v=O6yeLNNVa4A
Thank you for this article John. Perhaps useful to fellow writers, the Authors Guild has launched a Human Authored Certification program. If you are a guild member you can get the certification on your books / works free, but it’s a modest fee for non-members.
https://authorsguild.org/human-authored/
The artisans, craftsman and workers who know how to do and make things from their own imagination will be in need!
On another note, as all of this decline and chaos continues to accelerate the second religiosity, I have looked at the hippie Jesus Freak group, The Trees Community in my latest article. Unlike other Jesus People groups of the seventies, after they came down from the acid, they got involved in liturgical Christianity at the Cathedral of St. John in NYC where Edward Nason West, spiritual advisor to Madeline L’Engle, became their abbot when they took formal vows and created a monastic community. L’Engle was also the “mother” to this group as West was their “father.” James Parks Morton also played a role, and through all this, I look at the roots of West and L’Engle’s universalism, and Morton’s work in the interfaith movement and what effect this all had on the group, who already been explorers of the worlds religions before their conversion. I contrast that with the Jesus People who went the way of premillenial dispensationalism. I really love the music of the Trees Community, wanted to share their alternate tour through the realm of Christianity with others.
Interested readers can check it out here:
https://www.sothismedias.com/home/freaking-out-with-jesus-in-the-trees-community
and here:
https://justinpatrickmoore.substack.com/p/freaking-out-with-jesus-in-the-trees
JMG,
Your post is so understated, so well put, so spot on, so true. Thank you for this one!
JCR
Hi John,
Great post. I have a few comments to make:
1) In regard to white collar work – and in particular dysfunctional public service bureaucracies – I wholly agree that much of what they do is irrelevant or downright negative for what remains of our productive economy.
This is a UK example today but what is clearly a smart guy got so disillusioned with the utter incompetence of the British state he has decided to resign. The Foreign Office isn’t interested anymore in actually implementing and executing effectively whatever the British interest is but rather is woke, dysfunctional and ineffective.
https://www.thetimes.com/article/uk-foreign-policy-is-choked-by-lawyers-says-departing-diplomat-vpg22fpmz
I have personal knowledge from people close to me who work in government just how useless most civil servants are. Scope for massive reductions are clear to me.
2) Now, AI is an interesting one. I do use AI apps and they have improved their capabilities over the last 6 months or so. Overall logic and reasoning have improved a lot and where their data set is limited and high quality they make things far more effective.
Examples include using a regulatory chatbot to search their local rules and guidance. Before I would need to try and locate a specific clause on the website, now I can use the chatbot. I have also used Gemini AI for some personal research recently on areas that I have extensive knowledge over. I was quite impressed in parts and at its best its like chatting to a very clever friend.
However, I have also experimented with it in areas where I specifically ask for things (e.g. events locally) or what is going on in that country and it hallucinates things. A silly example was I had a spare 30 minutes and asked AI for local art exhibitions. It told me about one on landscape art. I didn’t check it and went to the art centre and it turned out that this exhibition is only starting in April! Not a big deal but this specific hallucination shows the fundamental weaknesses of the AI.
Yes, it had picked up an exhibition but couldn’t tell me that its not on at the moment.
I see a future with AI on AI with very specific and clean data points and/or AI tools that can find pattern recognition and reasoning leaps in industries with huge amounts of data to be tapped. Finance, healthcare, coding, insurance and so on.
Within the limitations of AI I’m sure some AI agentics will work and help automate tasks done by humans to now.
However, my playing around with AI tools hasn’t convinced me it can take over the really challenging and complex work done by white collar workers.
A couple problems more critical than your (correct) observation that LLMs won’t ever be able to achieve what industry hype has been claiming. The industry’s business model totally depends on achieving human-like AGI by 2028, which is the only way it can achieve trillion-dollar magnitude revenues increases needed to repay trillion-dollar magnitude data center capex spending. Previous bubbles (1870s railroad, fiber-optics 2008 housing) involved mistimed/premature investments in proven technologies, while the LLM bubble is a much larger one-way bet on a non-existent technology (imminent AGI). Add to this an even bigger political bubble. Every major political interest—the Tech Industry, the financial industry and the entire Washington establishment—has placed a one-way bet that an AGI-driven productivity boom will be the magic bullet that drives their desired programs (increased Tech Industry political power, equity markets “numbers always-go-up”, funding for lower taxes, higher military spending, etc). No previous bubble had this powerful array of forces dedicated to preventing the bubble from bursting and avoiding any accountability for the massive misallocation of capital.
Fair to speculate about hypothetical future impacts on different jobs/industries from LLMs, but the path from here to there is likely to be a lot uglier than your post suggest.
“Understanding the LLM Bubble” American Affairs Journal, Spring 2026, p. 122.
https://horanaviation.com/publications-llm%2Fuber
Synchronicity has struck, here is the opposing viewpoint.
“At Mobile World Congress, Cristiano Amon of Qualcomm argued that the coming 6G networks will power an AI-driven “agent economy,” where devices and AI assistants constantly communicate across the network. “AI will fundamentally change our mobile experiences,” Qualcomm chief executive, Cristiano Amon says. “It’s going to change how we think about our smartphones. Think about our personal computing. Think about and interact with a car. The car is now a computing surface. If you actually believe in the AI revolution, 6G will be required. Resistance is futile.””
“The traffic that we expect on 6G is way different than what we had before,” says Palkhiwala. “Before, it was all about consumer traffic. We expect 6G to be driven by [AI] agent traffic. Think about all these use cases where there are AI agents sitting on various devices — your glasses, your watch, your phone, your PC. These agents are going to be talking back and forth across the network to other agents and services. “The traffic completely changes. 6G is being built with this idea that the traffic that goes on the network is not just going to be consumer voice calls or downloading videos, we’re going to have agents talking to each other, so the reliability of the network becomes very important.”
https://tech.slashdot.org/story/26/03/04/0033216/qualcomm-ceo-resistance-is-futile-as-6g-mobile-revolution-approaches
I’ll be giggling about that for the rest of the day.
And then there was this lament from the Administrative State; “Dec 24, 2025 — With fewer than 40 bills signed into law as of Monday, the House and Senate set a modern record for lowest legislative output in the first year ”
The full article in the Washington Post is paywalled, but if you google fewest new laws passed you’ll find similar laments.
Tony A,
The cloud is unsustainable in the long run. It is hugely vulnerable to disruptions such as war, with Iran bombing AWS cloud data centers and taking down the entire AWS Middle East region with significant negative effects on businesses in that region being a very recent example.
I have to give credit to a YouTuber, Whatifalthist, actual name Rudyard Lynch, for putting a bug in my head to think about wrt catabolic collapse. He said something along the lines that after World War II the West in general was burning through its “social capital.” This would be things like the old unspoken social contract that a university would not only provide you an education, but also almost guarantee that you would be able to find a job after graduation that would bring you into the middle class. Or that divorce is really high these days in that both parties do have some incentive to defect from the marriage and there is little community or family cohesion that would encourage a couple to stay together. Basically, its the transformation of a high trust society to a distrustful society. Curious where this fits into the scheme of catabolic collapse.
Also, the thing I found that will correct the thin, tinny sound of streaming music is either to buy an instrument and make your own music or go to a live show. No digital compression loss there!
I was a kid when CDs came out, a teen when they really caught on. Records played on a halfway descent rig have always sounded better than CDs and most audiophiles I know agree. What I remember at the time that CDs caught on was that most of us late gen Xers were listening to audio cassettes at the time, which sounded horrible, and deteriorated over time rather quickly (I remember it being a badge of honor to the fandom of a band to have worn out your cassette, but I digress), so when CDs came out they sounded like progress incarnate compared to cassettes which is what my peer group were comparing them to, but scratched easily and at least at first played poorly on smaller players.
I went back to vinyl in my mid to late 20s right when the mp3/iTunes era had just started. I was at my mother’s house and found an old record of mine from childhood. Threw it on her player and realized records sounded better than anything that’s come out since, and if cared for lasted a lifetime and more. My mother has since passed, but I have about 100 of her 45’s from the 50s and 60s and most of them still sound great! Only problem in the last few years is that due to folks starting to go back used record prices (mostly classic rock stuff) has skyrocketed, but isn’t too bad, still much much cheaper than buying stuff new digitally. There’s a lot of smaller record presses coming out with real high quality/fidelity stuff as well, it’s a really cool time for vinyl.
need I cite Hollywood, which has completely lost the ability to make movies that audiences can enjoy?
****
They still can make movies that people could enjoy. Too many, however, are only making movies for themselves and their little clique of admirers. They’re contributing to catabolization in that they are using resources that could be making money. This eventually leads to fewer productions going forward due to lack of returns from the previous productions.
John,
Thanks for this great wake-up call of an essay. In January, I graduated from the bureaucrat factory (college) and got my job at a bureaucratic organization. The first thing I’ve learned is that I hate the cubicle. I got my college degree at the insistence of my parents and went the route of least resistance. Now I realize that working indoors is just not what I want to do.
I have experience working as a landscaper and for an arborist. I’m going to look into the trades and see what I can do. I enjoy working outside in the sun and the trees.
Tony, thanks for this. These are points worth considering.
CC, yes, I’m familiar with that claim, which was pushed heavily by the mass media while CDs were replacing vinyl records. May I offer a counterpoint? I have a stereo system that will play records, cassette tapes, and CDs, and I listen to all three. Analog recording has a richness of sound that digital recording cannot match. No, it’s not just “crackles;” it’s a distinct qualitative difference, noticeable as a “flat” quality on digital media. What I’ve noticed, though, is that not everyone seems to be able to register it; I’m not sure if people like you don’t hear it or simply have learned not to notice it. That said, for those of us who do perceive it, it’s a major difference — and it explains why vinyl records are resurgent nowadays.
Paul, human bureaucrats have multiple motives for increasing the regulatory burden — it makes them feel like they’re accomplishing something, it justifies hiring more subordinates, and it sharply increases the potential for bribery. LLMs are not subject to these motivations. I freely grant that the people who run then will be just as corrupt as their bureaucratic predecessors, but if it’s a matter of bribing one employee rather than a whole office building of them, the burden is noticeably less.
Richard, what’s already happening, of course, is that businesses are abandoning overregulated states. Smaller firms are moving elsewhere, larger firms are simply shutting down their California stores and offices, because it isn’t worth their while to do business there. California in particular is making a cascade of massive mistakes that will turn it into the Rust Belt of the 21st century; I confidently expect that by 2050 or so, SF and LA will look like Detroit, complete with much less than half their current population.
Patrick, ah, but it’s easy to tell a nonfiction book written by an LLM from one written by a human being. My guess is that LLM books will turn into a major money-loser, and be abandoned before most other LLM products.
Other Owen, keep snorting away! 😉
Justin, interesting. I may need to give the Authors Guild another look.
John, thank you, but we’ll see if it’s spot on as things unfold.
Forecasting, as I noted in the post, I think there are definitely many things that a well-trained stochastic parrot can do, and do profitably. Replacing overpaid bureaucrats is certainly one of them! The problem, as with the internet bubble that popped in 1999-2000, is that a genuine advance is being overinflated to the point of absurdity.
Hubert, fascinating. That makes sense, and it feeds into my take, still tentative but developing, that we’re going to see a serious economic and political crisis across the Western industrial nations by the end of this decade. If everyone in power is counting on humanlike AGI as their “Hail Mary” strategy to pull out of the power dive, that becomes even more likely. Thank you for the link, btw — if you know of other resources that someone without a tech background can use to follow up on this, I’d appreciate hearing about them.
Siliconguy, thank you. I needed a belly laugh. As for Congress, good. Lots of new laws are not necessarily an advantage!
Watchflinger, excellent. Social structures are also capital, and we spent the second half of the 20th century subjecting them to ruthless catabolism. As for live music, well, that’s even better!
Dean, oh, I know. My stereo has a cassette tape deck but I use it only when I can’t find something in any other format. As for vinyl, it’s always a crapshoot, but many thrift stores have record sections where you can find astonishing things for reasonable prices. I can also point you to a great used record store here in Silver Spring…
Clarence, they could theoretically do so, but they’ve forgotten how. The quality of acting has also gone sharply downhill in recent decades. Watch the acting in an old classic like Casablanca sometime and compare it to what passes for acting these days; the deterioration in technical skill is very striking.
Nephite, all I can say is “move fast.” If you send your parents some articles about how AI is going to replace all the cubicle inmates, they might not argue as much.
John,
I have a lot of connections from my old landscaping job in my city’s Parks and Recreation department and it would be pretty easy to go back. Since it is a city department, is that too bureaucratic to be secure? I’m wondering whether to go back to that or to work for a small business like an arborist, xeriscaper, or private landscaper.
CS Lewis ‘Pilgrim’s Regress’ has a chapter called ‘Parrot Disease’. After Reason defeats the giant and opens the dungeon door so his prisoners can be free, they refuse. She gives up and rides off, remarking ‘This psittacosis is a very obstinate disorder’.
I really hope you are right that AI will be the beard for a serious attack on US D party pork. A three-way fight between pork barrel oligarchs, financial oligarchs, and tech oligarchs is the last best hope for the rest of us.
A great article here from Ludovic Viger who is a reader and sometimes commenter here:
https://thegreatcanadianreset.substack.com/p/localism-isnt-romanticits-structurally
It’s localism and appropriate scale of systems. I think it relates very well to the discussion of bureaucracy.
“apartment blocks twenty stories high in 300 AD”
Surely nine storeys, and after the time of Augustus, seven stories? Without elevators, those living in twenty storey buildings would have a very hard time indeed.
Hello JMG and commentariat:
I’m glad to some extent you’ve written a new post, less related with actual news than last Open Post (there’s some relief of hot arguments about them IMHO). However, after a first fast reading, it seems to me it has a strong political aspect, in the good sense of that term.
I’m thinking now about your ideas, John, so in not very much time I’ll write my opinion about them.
My comment is NOT related to the post at hand.
Now that you’ve moved to the DC Metro area, you’re within a few hours by train from Harrisburg, PA.
In Harrisburg is the Midtown Scholar, a GIGANTIC, scholarly used book store. They must have several hundred thousand titles in their main store, plus more in the warehouse.
https://www.midtownscholar.com/
The Midtown Scholar is what its name says. They don’t carry mass-market genre paperbacks. Or even much genre fiction at all, other than classics and Today’s Literature.
Eric Papenfuse, the owner, just announced the purchase of a large selection of rare occult books from a deceased collector.
I have no idea how are any of them are but you might, eventually, want to take a look.
The shop is amazing. It gets bigger as you walk inside; room after room, level after level. I think he owns the entire block and has filled all the buildings with books. Plus his warehouse.
One aspect of bureaucratic labor I haven’t seen discussed is the fact that bureaucrats are disproportionally women. Hugely so from what I see. When the layoffs hit, we’re going to see a startling level of unemployment in PMC women. This demographic has a quite a level of student loan debt from the degrees they earned to obtain the job they just lost. A significant fraction of those who are single bought houses, and they all bought nice cars on loan, and commonly carry credit card debt.
When the country de-industrialized in the 80’s, it largely impacted men, and those jobless men commonly lost their wives or girlfriends next.
I’m sure the married former bureaucrats will be the best off, as I can’t imagine an otherwise content husband would discard his wife over her joblessness. I can only guess what this will do to feminist ideology and culture. I can’t fathom what this will do to the dating market, and relations between men and women generally. One might think that single men will benefit, but something tells me this will not be the case; I can’t put my finger on it. We are certainly entering uncharted territory.
–Lunar Apprentice
(Off topic) Anselmo: I’ve just read your last comment (# 515) in the last Open Post. I can’t answer you there because it’s now closed (I’m sorry). In the short form: No argument here. I agree. Saludos.
Wow, you really hammered that stake home through the vampire, the coffin, and the crypt floor. Hard to deny, now, maybe impossible. It occurs to me that there is simply (also) the cost of people who MIGHT be doing something productive, not doing it at all. Maybe that’s taken in, in the enumeration of the other costs. Still, if you pull half the workforce out of labor pools and into cubicles, you have an enormous opportunity cost, if all those cubicles are already a net drain: now they are sinking alternative opportunity costs, along with all the active waste. Incredible.
I kinda wrote about this over at my place yesterday. Her is my money quote:
But my initial thoughts are that these things are not “intelligent” per se, what they are is the scheme for replacement of the middle layers of a system of production that has gotten too large and was never intelligent in the first place.
But you have to remember that over 50% of employment is in “white collar” jobs that are in the crosshairs. These are embedded throughout the manufacturing, local government, education and service sectors and my guess is that they will start getting nasty when things start going against them, and the losses of income and “prestige” of this particular flavor of critter will cause a smallish minority of them to generate the next external proletariat.
Fun times ahead.
Good post JMG, my observations:
Although I am retired now, in my career I worked in both private and in public sectors. In both, the people who either brought in the money (private sector) or directly served the public (public sector) were often subservient to non productive people who did not directly serve people paying the bills. Early in my career, I am a geologist, I could escape these pests by doing fieldwork, but with the “progress” in communication technology, my autonomy was severely curtailed by busybodies giving “guidance”.
I feel sorry for young people who will never no the joy of working without someone looking over their shoulder.
There’s another qualitative element to the vinyl resurgence beyond the superior audio quality. Relaxing in the evening, putting on a record, pulling it out of its sleeve where you can see the artwork much better than on a file or CD case, listening becomes tactile. You turn over the record, and you pay attention. You slow down. All of these are just as vital to the experience and part of the allure.
I actually acquired a new walkman so I can start listening to tapes at work again.
I had a good time with my own table of used music, used books, and new by-me books (and a few other titles) at the local twice yearly record fair. I made a little bit of money. … I am not sure I want to do that kind of thing all the time, but my grandpa raised a family of five as a newspaper boy and having his own newstand from age 14 to 75.
I thought of something similar, but there would need to be newspapers and magazines worth selling.
A used bookstore might be another possibility, but don’t know how viable it is… they tend to be my favorite.
Learning some freelance writing skills is definitely on the list for other businesses though, among other things. I have a bit of food prep knowledge, have catered with my wife on many occassions. As we continue to age, and she is older than me, she’s not telling people about it though, because it is serious hard work, as is the restauraunt biz… want to find something more sustainable… some kind of speak easy on occassion to make some dough would be cool, and is something we’ve talked about.
I will continue to brainstorm though.
Here is my book, record, CD and tapes stand from the record fair. Since I was in the basement and brought my own table I didn’t have to pay the organizer as much as the big dealers upstairs:
https://substackcdn.com/image/fetch/$s_!UfT3!,w_281,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fce9c7bb3-e6da-42b8-a1ab-aafc80289789_4000x3000.jpeg
https://substackcdn.com/image/fetch/$s_!dDW-!,w_281,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb28c4db7-bcfa-45fa-a52e-984f0649dc8c_4000x3000.jpeg
What businesses need is customers with money in their pockets. So the business cycle will become the business death spiral.
The first round of layoffs will be due to AI replacing employees.
The second round of layoffs will be due to fewer customers due to the first round.
The third round of layoffs will be due to even fewer customers due to the second round.
Continue to the sound of water gurgling down the drain . . .
Just last week this article about AI decimating white collar jobs was making its way around Twitter:
https://www.citriniresearch.com/p/2028gic
A Chinese writer wrote this response/parody:
https://www.eastisread.com/p/the-2028-chinese-intelligence-crisis basically saying that all the things AI can do don’t really do much in Chinese society and economy.
I think both are worth reading, the Citrini article puts in writing what I’ve been saying IRL. I feel like both parties overstate their case but China definitely has a headstart, Xi already cut the private tutoring sector, the real estate sector, along with the associated financial sector in China, all avenues for graduate careers that don’t directly contribute to production. He wants people to retrain in new technologies and new manufacturing instead.
I think there’s another factor at work with the AI bubble. I think one driving factor is a desperate effort by the tech companies to avoid losing the unknown, but very large, amount of resources and skills that they have invested in neural networks. Neural networks are an old technology: the first functional example is the Perceptron from 1958, which could learn to identify letters and numbers. This sounds really simple, but it is actually a major accomplishment; it would require a fairly substantial amount of space to explain why, but this is actually incredibly hard to code.
They faced a few problems, but one of the most important has been that they are extremely computationally expensive. As before, I’ll spare the details, but they are one of the most computationally expensive type of software that there is. Even worse, because it’s remarkably hard to figure out what’s happening internally, there’s no real way to optimize the code. This is a huge problem, because it means that they remain computationally expensive to run; which means that they require a lot of infrastructure, ranging from the hardware to run them on to the massive amounts of power to run the hardware. This meant that there wasn’t a lot of research going into them, because everyone knew (correctly, I think) that even if they were viable as a technology, the computational requirements were so intense that they would never be viable economically.
In 2012, this changed when a neural network known as AlexNet won an image classification contest, ImageNet, in a landslide. AlexNet was computationally intense, but less so than most people thought would be needed, it defeated a lot of programs which were also computationally expensive, and it did all of this with code that was actually pretty simple to set up. This seemed to suggest that neural networks might actually be viable after all, because there was a lot of potential in something that could classify images, even if it was computationally expensive. Add to this the general sense in the computing world that computation always got cheaper, and the result is that a massive boom in neural networks occurred.
For a little while, this seemed to work. What a lot of people involved missed though was that each advance required more and more computation; and at this point, the total amount of computation that is required to run these systems, to say nothing of training new ones, is a sizable fraction of the total costs associated with running the tech companies. These costs are large enough that in most cases these systems cannot be run on a local device, and so in order to run them at all in any place where people are, some level of internet connection is required. This is why so many modern devices and software cannot function without the internet: some part of it has been handed over to a neural network, and that part of the program won’t fit on the device.
For a while, this seemed to work well enough: computation did indeed get cheaper, for the same amount of computation. This ignored the fact that neural networks were so computationally intensive that a lot of companies started paying more total computational costs; and while for about a decade or so after 2012, this seemed to pay off, it looks to me like we are now past the limit where scaling the networks makes them more functional. The problem is that right now, it’s hard to come up with a use for them that justifies their intense computational costs. While if they were still scaling up, people could imagine AGI, which would be well worth nearly any computational cost, I think it’s pretty clear now that neural networks can’t get there, and so the tech companies are left with these bloated, monstrously large and expensive programs.
The catastrophic misallocation of resources is even worse than the above would suggest, because neural networks are best run on what are called GPUs, which is short for Graphical Processing Units, because they were initially designed as specialized chips for graphics. The key point though is that these are specialized, and not general purpose, chips. This means that if it turns out that neural networks are not useful on a massive scale, a very large amount of the investment is going to turn out to be wasted. There really are not that many known uses for GPUs outside of graphics and neural networks; so if neural networks are useless, then all the GPUs that major companies bought are worthless.
Even worse is this applies to all the infrastructure built to get these chips: all the manufacturing for GPUs, all the massive infrastructure to secure a supply of them, all of the massive training that the tech companies have done, all of it will become worthless if it turns out that there is not something that these bloated and overly complicated neural networks can do. I suspect that a huge part of the overblown claims and efforts to put neural networks
Given all this, I think there’s a high likelihood that this AI boom is part of desperate effort to find something, anything, that they can get these giant systems running neural networks to do. The alternative is writing off a huge fraction of their assets, and I suspect it would be enough to drive a lot of these companies into bankruptcy, which would instantly be a major economic and political crisis. It might even topple governments, because the tech companies are the “Wave of the future”, and seeing them fail in spectacularly messy fashion might break the myth of progress, with effects that are probably impossible to predict.
The Supreme Court has let lower court rulings stand denying copyright protections to AI generated art. I presume the same is true for AI generated texts.
As for bureaucratic reform, ala DOGE and other sweeping cut-backs, it’d be nice if they used “Chesterton’s fence” as a guideline, rather than tearing up things with no care for why they were originally imposed. But such careful scrutiny seems improbable by anyone these days.
Funny I should read this immediately after Musa Al-Gharbi’s article today about the economic foundations of the ideology of symbolic capitalists – which is to say, why the professional managerial class (PMC) are woke. Simply, to mystify self-serving bureaucracy. (So successfully that I suspect most don’t realize this is what they’re doing )
“The U.S. collects less in taxes relative to GDP than most other OECD countries, but this is not because the wealthy and corporations pay a lot less here than they do elsewhere (in fact, they pay more) — it’s because everyone else contributes much less in income and payroll taxes in the U.S. than elsewhere.”
“American policy . . . has been consistently oriented around taking more narrowly from the rich and giving more narrowly to the poor. . . . money gets taken from the rich and funneled into institutions and programs that symbolic capitalists control, and we eat the vast majority of the income that passes through our hands, and then sprinkle the remainder upon the disadvantaged once we’ve had our fill.”
Al-Gharbi compares the role of the PMC to the original story of Robin Hood, wherein Robin stole from the rich and… kept it for himself. The PMC also become gatekeepers with power over the lives of others. In the final footnote he provides evidence that this leads to loss of faith in institutions. Much more detail in the article.
https://musaalgharbi.substack.com/p/on-redistribution
Very interesting essay JMG, as was the fact that the essay below was next to yours in my inbox today from Readup…Thank you for a fascinating morning read.
“AI companies will fail. We can salvage something from the wreckage | Cory Doctorow
https://www.theguardian.com/us-news/ng-interactive/2026/jan/18/tech-ai-bubble-burst-reverse-centaur
So do our overflowing bureaucratic office fauna comprise a broader family or order or class of civilizational dross? I don’t think they could be a genus, because one of their most abundant subgenera would be the common parrot, which then contains a bewildering variety of species such as stochastic, dead, grumpy, etc.
Could it be that LLM’s are just a strange new variety of bureaucratic office fauna, one that doesn’t continuously demand an input of costly crackers? They’re so much more easily satisfied by ever bigger data centers. What a waste that we have to build such gargantuan false idols just to be able to feed our society’s crippling dead weight into them.
JMG and Nephite/Neophyte,
I hire a local sprinkler/landscape guy and a local arborist as needed. Both are former corporate managers with degrees to match. Both decided several years ago that no, that was not how they wanted to earn a living. And, both are still available after all of these years to do the little jobs I want skilled people to perform on my little patch of suburban land. This tells me that they are doing just fine, thank you very much.
I thought about sending this excellent essay to a PMC family member, and decided against it. They have a recently college-graduated nephew with a marketing degree who is underemployed at a pickleball business. His mom sends him money to make up the difference between his paycheck and his desire to spend the balance of his workday playing video games with his buddies. IMHO not a good long term plan, but truly none of my business. I’m already regarded as the family loon.
OtterGirl
Siliconguy @ 11: ” “It’s going to change how we think about our smartphones. Think about our personal computing. Think about and interact with a car. The car is now a computing surface.””
I believe that the cars of the future will become so advanced that they will have no moving parts. Meanwhile I enjoy my stick-shift Fusion.
In my news feed on Monday: “AI used nuclear weapons in 95% of war game simulations, study finds.” https://ground.news/article/ai-used-nuclear-weapons-in-95-of-war-game-simulations-study-finds
Who knew that those 80s movies like Terminator and Robocop were actually prescient documentaries? I have to say, though, for people like me, who have to spend ages trying to formulate a search-engine query (or, back in the days of card-catalogues, hours of looking through index cards) to get a satisfactory result, the advent of natural-language queries has been a boon. The rest, of course, sounds like the universal snake-oil parodied in “Medicine Show” by Big Audio Dynamite.
Meanwhile, any talk of cutting bureaucracy or regulations gets greeted by the Dippers and other professional Social Activists with shrieks of rage, indignation, and fear about what terrible apocalypse would befall that abstract concept they refer to as “The Poor”. Despite the decades of policies based on that same “caring desperately about the poor” philosophy put into place in Toronto, and across Canada, the numbers have just steadily climbed. Shirkey’s Law in action. (Or inaction, perhaps?) One might suggest it is time for a different approach? But Bureaucrats change slowly if at all.
Cutting regulations so manufacturing and extraction industries can again function as engines of wealth (if not growth) inevitably produces similarly hysterical polemics about ‘saving nature’ from rapacious, destructive Evilcorp &c. Or, since that tactic is now failing, the reason not to change anything now focuses on UNDRIP and our being historically unprecedented evil White People. (As an aside, it occurred to me that our intellectuals have gone from praising our special place as ‘Greatest Culture of All Time’ 100 years ago to ‘Most Evil People Ever Seen’ … which seems to me to be the same coin of arrogant self-importance, just flipped over.)
And yet… a recent poll shows most people (58%) agree with the aims and goals of President Trump, to cut back on government but the vast majority (90%) are appalled at the methods. Meanwhile, in Canada, we were much more polite and last year managed a net loss (‘degrowth’ — cf. ‘ungood’) in population for the first time ever as many people were quietly persuaded to head back to wherever they came from as we modified our immigration policies, and became far less indulgent towards people who overstayed their visas. No neo-NKVD required.
As for AI-proof jobs, my spouse and I are looking into a farm because we are at last able and ready to make the move. It’ll be a learning curve, for sure. But she has her personal business (AI is absolutely forbidden in the legal profession), and I will become a retiree able to unleash all my pent-up interests on the world. 🙂
Well, like I’ve written before, I think everything about AI and bureaucracy has a political meaning, in a wide sense.
John, I’ve also thought there was some relation between the current AI frenzy and bureaucracy before reading your current post, but I must recognize you’ve explained it in a better and a more provocative way than I could have said/written it in my own words.
AI “progression”, beyond today near evident hype, means to open a can of worms. Of course, quite bureaucratic jobs can disappear soon. AI freaks and promoters also love to say happily that even in a not far future, AI-robots will replace every doctor and surgeon within medical services (oops!). This wild optimism shouldn’t be believed literally, but it shows what are the tendences…at least in the wishes level.
I tend to distrust the AI madness not only by its evident risk of economical collapse (oh bubble…), but because I guess its most fierce supporters have (at least in a subsconscious level), some hatred against real democratic management. Indeed, if/when AI will replace bureaucrats, why not replace politicians too? Yes, nowadays politicians aren’t very good doing their job, methink…but maybe AI wouldn’t be better than them. I think there’s a real danger that more and more political decisions could be “robbed” to average people by the AI frenzy (unless its cards castle collapses soon). By the way, I tend to use AI as less as I can allow it for myself.
You’ve also written about USA bureaucracy as the biggest of the world and in historical terms, but I bet you that we’ve got in Europe a bureaucratic monster which could be at least the second in its size and power: the EU structure. A large bureaucratic and powerful machine which very few democratic controls (for example a Parlament with not much real power), ruled by non elected bureaucrats who decide about quite topics over the states members, especially in the economies. It’s paradoxical EU rulers love to preach and compel Europeans into “neoliberal” politics…but Brussels is in itself a para-statal bureaucracy whose budgets have never dwindled (ahem). What a contradiction…
One problem with stochastic parrots replacing bureaucrats is that there is literally no way to stop them from BSing– “hallucinating” in the speech of the toaster frakkers.
See: https://hackaday.com/2025/10/10/your-llm-wont-stop-lying-any-time-soon/
Now, the people employed in public service aren’t infallible, but they’re much, much less likley to make up rules and regulations out of whole cloth and try and apply them to you and only you. The state has to care about that because it creates a liability: that is, you can sue them. I’m pretty sure the TOS says the LLM vendor is released of liability, so it’s on the org employing the thing.
So far I know the UK has studied this and found Microsoft CoPilot completely useless, not helping with their bureaucrats productivity in the least. Having talked to people who’ve been encouraged to use stocastic parrots in office jobs, they often start out very enthusiastic and end up frustrated and disillusioned.
When you talk about using the LLMs as and excuse to downsize and pare back the bureaucratic state — yes, I can buy that. If you think they can prop said state apparatus up while using fewer resources than the human paper-pushers they’re supposed to replace, I think you are both overoptimistic about the technology and underestimating the full resource costs, which… is not something I thought I would ever say to John Michael Greer.
Nephite, city jobs that actually do hands-on work are likely to do tolerably well, at least for the near to middle future, and that will give you time to start getting ready to hang out your own shingle.
Bruce, oh, it won’t just be Democratic pork on the menu. It’s a bipartisan swine roast, and some parts of the GOP will be providing plenty of ribs for barbecuing.
Justin, excellent! He’s right, of course — one of the most effective ways to catabolize excess bureaucracy is to decentralize, and leave much more in the hands of county and state officials or (quel choc!) ordinary people.
AA, interesting. I looked this up and find that I was misinformed. Correction incoming.
Chuaquin, thank you. Not everyone takes the time to think…
Teresa, hmm! It’s not a day trip — I just glanced at the schedules, and it would be around six hours each way via train — but I’ll keep it in mind.
Lunar, good heavens. I should have taken that into account, and didn’t. Yeah, that’s going to cause some dramatic changes.
Celadon, that’s a good point. Among the capital that will be freed up for productive uses will be employees, though most of them will need retraining.
Degringolade, exactly.
Raymond, there’s that! It’ll be interesting to see if there’s any significant trimming of busybodies.
Justin, used book stores rarely break even. I’d suggest writing — you have a good grasp of style, and I can tell you from personal experience that it’s possible to make a decent living that way…provided that you work like a dog. Get at it!
Martin, ah, but there’s a flipside. As the global economy disintegrates and the bureaucratic barriers to small business creation come down, there’ll be much more call for the production of goods and services locally — and those can and will provide paychecks. At first there will be a very sharp downturn, no question, but it’ll bottom out and recovery will begin, though to a lower final level. More on this as we proceed.
Alvin, thanks for both of these. I’m watching China closely at this point.
William, fascinating. Thanks for this! I wasn’t aware of the details. The sunk costs fallacy claims another round of victims…
Bill, I haven’t heard yet about LLM-generated texts, but here’s hoping.
Alphonse, al-Gharbi’s one very sharp thinker, and I think he’s on to something important here.
Jill, and Doctorow’s probably right. It would not surprise me a bit if ChatGPT is the next Pets.com.
Christophe, I’ll leave classifying the various species of dead parrots to the necropsittacologists!
OtterGirl, good to hear about the former bureaucrats. As for the soon to be former bureaucrat, the fact that Mom’s keeping his standard of living at an overinflated level does not suggest any great willingness to face reality…
Renaissance, best wishes for the transition to farming! It strikes me as a smart move.
Chuaquin, I meant to include European bureaucracies — it’s Western industrial civilization as a whole that has a historically unprecedented case of metastatic bureaucracy. Surgery is called for.
One aspect of this i find encouraging is that because, sadly, 1 in 4 Canucks “work” for the gov’t. Either municipal, provincial, or federally. A possible civil war does not necessarily require blood to achieve change and removal of these grossly incompetent wankers. The deterioration ( especially in BC) will accelerate the Balkanization and, hopefully, change for the better for my offspring and their wee barnes. A excellent summary of just how bad things are in Canuckistan at John Carter’s postcards from barsoom.substack.com
Warning to my fellow Canadians though it’s a tough reading.
John, I appreciate the piece and largely agree with the bureaucratic collapse thesis. It’s probably the most underappreciated macro story of the decade. But your framing of LLMs as stochastic parrots, while technically defensible in a narrow sense, leads you to a conclusion that’s too uniform. And I say that as someone with skin in the game on both sides of this argument.
For context: I’ve spent years working with AI teams at top-20 pharma companies, where these tools are being used for predictive modelling in drug discovery, protein structure analysis, and genomic research. Today I run a business that builds AI workflows for SMBs. I use these tools every single day, at depth, across wildly different domains. That vantage point matters here.
You’re right that the hype is detached from reality. The consumer-facing AI narrative is inflated, and the data center buildout has the fingerprints of dot-com malinvestment all over it. But “the hype is wrong” and “the technology is useless” are not the same claim, and I think the piece blurs that line.
The use cases are not consistent, and that inconsistency matters enormously. Some applications are under-hyped while the mainstream conversation fixates on the overblown ones. Frontier models are contributing to genuine scientific discoveries that get almost no mainstream coverage. The gap between what people think AI does and what it actually does at the frontier is as large as the gap between internet hype in 1999 and what the internet actually became by 2010.
The mundane case is also being underestimated. A document that would have taken me ten to twenty hours at moderate quality now takes thirty minutes at higher quality. That’s not a parrot. That’s a leverage multiplier on human cognition. The implication isn’t wholesale replacement of humans. It’s that what previously required a team of twenty can now be directed by one person with good judgment. You don’t need full replacement to fundamentally restructure labor economics.
The deeper issue is that you’re conflating LLMs as they exist today with the trajectory of what they’re enabling. Every major frontier lab is using AI coding and research tools to accelerate their own development velocity by an order of magnitude. The models being built on top of LLM infrastructure use explicit reasoning architectures that address the exact limitations you correctly identify. LLMs aren’t AGI and won’t be. But they’ve provided the scaffolding, in accelerated code generation, scientific literature synthesis, and physics-realistic simulation environments for training robotics models, that brings genuine machine reasoning meaningfully closer and faster than almost anyone outside the field appreciates.
The bureaucratic collapse thesis is compelling and stands on its own. But bundling it with a sweeping dismissal of the underlying technology is worth a second look. The dotcom era had its share of sober analysts who were completely right that Pets.com was a joke and completely wrong about what the internet would become. Being right about the bubble doesn’t make you right about the technology. I’d genuinely encourage digging into some of the frontier use cases before drawing the broader conclusion. The view from 30,000 feet, looking only at the consumer hype cycle, is not representative of what’s actually happening at the edges. That’s where this gets decided. And right now, that story is barely being told.
I’m reminded of Larry Niven’s short story, The Schumann Computer. As a practical joke, aliens give a human the plans for a super intelligent computer. It costs a fortune to make and once operating it demands vast amounts of information and sensory data before it will answer any questions. And once it has all the information, it shuts itself off.
Grok is a modestly competent search engine if you’re careful about the queries you make, and check for yourself to make sure it’s not lying. I don’t have conversations with it anymore.
I’ve been aware that this system has to collapse since the early 90’s. And I’ve been watching the process since then. Yet tomorrow is another day.
My wife drags me to live music tolerably often. I do enjoy it more than prerecorded.
Thanks again for another thought provoking article.
JMG et al,
While I agree that over regulation is bad (says it right there on the tin) I worry that safety regulations are the babies getting thrown out with the bath water. Whether it’s environmental protections or OSHA or any of the thousands of facets of running Industrial Civilization that require safety rails such as transporting chemicals, oil and gas, or nuclear materials between facilities and on freeways, how to properly handle and store said materials, safety measures to protect workers in factories, manufacturing centers and construction sites, what can be safely used to make, treat and process our food from farm all the way across the vast agribusiness network to a far distant table, at every step there are measures meant to keep the workers and the general public safe. I fear that along with a great many useless and frivolous regulations, factories owners will be quietly celebrating the end of “time wasting” and “inefficient” things like fire protections or redundant systems that cost extra so the replaceable cogs, sorry, “people”, don’t get maimed or killed by the profit motive.
Am I worrying over nothing? I’m having trouble finding the third path out of all these binaries these days.
I wonder what the world record is for the number of layers of government? In Wales, there are parish or town councils in most areas with each covering a population ranging from a thousand or so up to a few tens of thousands. Above them are unitary authorities responsible for one to two hundred-thousand and then the Welsh Senedd which has some powers over the whole of Wales, a population of 3 million. However, most powers are still in the hands of the UK government in London. Before the UK left the EU, there was a fifth layer in the European Parliament. While parish councillors get no pay, meet in rooms in village halls, have almost no staff support, a tiny budget and hardly any power, all the other levels have abundant paid staff and dedicated offices. One unitary council office block built in the 1980’s was so grandiose cynical locals christened it the Taj Mahal!
In England, there are three local levels in most areas, parish, district and county though a process is underway to combine the latter two. This process is apparently so complex it is forecast to take at least five years. I will be surprised if the total number of staff employed after the end of this reorganisation is smaller rather than larger.
For the JMG and the commentariat,
Any recommendations on a good record player? It looks like the market is saturated with possibilities but I’m also finding a lot of “actually Bluetooth speakers and not record players” types too.
Part of my work is data engineering. I’ve dabbled in using AI, and it seems pretty unsatisfactory to me thus far. I can use it for small things such as writing regex. But designing anything complex….no.
But, let’s assume for a second that it does what it says on the tin. As a civilisation, do we a) use this expensive resource wisely (which consumes vast amounts of resource), or do we b) use it to replace customer service staff and to make videos?
Back to reality. This is my prediction: the circular funding scam collapses. The US economy is at risk, so Trump answers the call of his oligarchs – after all, they own him – and pumps billions of public funds into the market to prop it up, inflating the bubble. (At least, the billions of public money he hasn’t embezzled off into his private ‘Board (Bored?) of Peace’ bank account. ) And then you get a bigger bang.
JMG,
Thank you for this insightful post. I read your paper on catabolic crisis/collapse when you posted it a year or so ago, and enjoyed it very much. I have also read other writers who talk about civilizational decline such as Arnold Toynbee (thank you for the recommendation!), who talks about “creative elites” becoming non-creative and resorting to force, and Peter Turchin, who talked about intra-elite conflict, wealth gaps, and state bankruptcy, Michael Hudson, who talks about the unwillingness of creditor/oligarchic elites to forgive unpayable debts forcing populations into debt servitude or to flee, and Ibn Khaldun who talks about elite decadence. It strikes me that each of their theories on civilizational decline, along with yours, about the (in) ability to maintain “stuff”, are not mutually exclusive. I have oversimplified these theories, of course, but I am wondering how you see these varying schools of thought complimenting (or contradicting) each other?
I always wonder how the Hawaii locals who complain so loudly about the PMC tourists ruining everything will react when the planes stop flying as often.
JMG,
One of the reasons that LLM’s will sweep away huge segments of the bureaucracy is that much of the work that occupies these cubicle dwellers has been created over the last 20 years to deal with meaningless regulations and laws.
Until my wife retired this summer, she managed a large wastewater utility. Over the 40 years since she entered this industry as a young engineer the number of administrative employees required to deal with nonsense is dramatic. Back in the early 90’s she managed the 3rd largest treatment plant in Oregon. She had 65 union plant workers, a single clerk/admin assistant and a single white collar lab manager.
Fast forward to the larger multi-plant utility she managed before retiring. Just to deal with the requirements of Oregons ” equal wage law” , discrimination law, etc she had a HR department of 6 people, a 4 person legal team, an 8 person communications department, an 8 person compliance department and on and on.
Since much of this jobs are busy work that is just a response to demands for busy work it is easily automated by LLM’s. No original thinking is needed, just blind obedience to a bewildering array of laws and regulations.
As someone else mentioned, nearly all this ” busy work” is done by college educated women. Her line staff of treatment plant operators, truck drivers, engineers, and construction workers were almost all men.
Thank you John. It means a lot…
One of the reasons I joined the Authors Guild was to learn more about the business side of things from their webinars, documents and forums. Also, when I had my first book published I used their resource on contracts to get the one I had been sent amended to something that was more in favor to my interests (out of print clauses and the like) so I was really grateful they had that. Members also can get future contracts looked at by their lawyers, so I really hope to use that service before too long.
I’ll be silent on my current specific plans, and other things that are close to ready for submission & pitching, but I am working.
Back to it.
Interesting side note: the Supreme Court has just let stand a lower court ruling that AI-created works cannot be copyrighted – they immediately enter the public domain.
Tony A.#1:
You’ve remembered how much stressed is South Korean society. I’m not an expert in that Asian country, but I’ve heard there’s a high alcoholism rate there and its birth rate is low (not like the Japanese one, but still low methink). It’s interesting to see in South Korea quite people has become Christian, a curious tendence within Asian countries.
I’ve got the impression (after having read/watched some Korean books and movies), that South Koreans have to some extent, a hidden (or not very hidden) tendence to be agressive, at least in their attitudes.
I think South Korean society isn’t a healthy nor happy society. Maybe the wry memory of Korean War and fear/hate against its Northern neighbor have influenced people minds from parents to children until today, though its relatively fast race toward hi tech industrialization may explain it too (in addition to postwar collective trauma). In addition to this, South Korean democracy after the Korean War was broken (If I’m right) by two dictatorial periods, the last one lasted until the ‘80s methink. Maybe a last reason to social trauma.
—————————-
JMG and someone of the commenters have written about the vinyl records/CDs topic. I’ve just remembered an argument I had with a friend in the last ‘80s. He said CD was the definitive progress in music, and vinyls were going to be forgotten soon. I said vinyls had a charming quality in their sound and aspect that CD never would reach.
Today, vinyl discs are made, sold and listened by a not despicable minority, and CDs (and movies DVD) have been in decline since last years ago…to give pass to the online music, listened in PCs and more often in smartphones. So I was partly right: vinyl has dwindled but it hasn’t disappeared yet.
JMG, thank you for revisiting catabolic collapse and clarifying the difference between crisis and collapse
I have a quibble with one of your statements. “I confidently expect that by 2050 or so, SF and LA will look like Detroit, complete with much less than half their current population.” I agree that the San Francisco Bay Area as whole is likely to lose half or more of its population by then and that parts of it will resemble Detroit.
San Francisco itself is a special case. Owing to being built on a peninsula, it is surrounded by water on three sides; the land to the south is not part of San Francisco. The City’s population today is fewer than 850,000. I looked up the population in 2000, about 775,000, and in 1950, about 773,000, with minor fluctuations between those years. A great deal of the housing that was built by 1950 is standing and in use. The public transportation that served the neighborhoods in 1950 (trolleys and buses) is maintained and in use.
In the Financial District, the newer skyscrapers and office buildings are nearly empty, high end retail near Union Square has already collapsed, the commercial real estate market is dire. BART, the multi-county commuter rail line, would be bankrupt if it were privately owned. There was an obvious housing bubble South of Market building tiny condos for young people commuting to Silicon Valley; I bet that’s done.
But the population of San Francisco has not even doubled since the 1920 census (507,000) more than a century ago. I don’t think the City is likely to lose half its population in the next quarter century. The population and infrastructure of the East Bay are more at risk from natural disasters (fire, a large quake on the Hayward Fault) than San Francisco is from climate change, earthquake, and sea level rise.
„A new study from Princeton has important implications for education, scientific discovery, mental health, and more (perhaps politics and even decisions about war?). Essentially anyone who uses a chatbot is at risk. Because what is shows is that sycophantic AI that serves as a personal echo chamber that can actually keep you from finding good ideas. ”
https://garymarcus.substack.com/p/breaking-sycophantic-ai-distorts
TylerA, it’s entirely possible that the whole thing will turn into a disaster — but that doesn’t change the fact that LLMs seem to be being geared up to replace a lot of bureaucrats. We’ll see.
Longsword, that’s a great example of bureaucratization gone metastatically insane. No nation that hasn’t gone full-on Marxist needs a quarter of its workforce employed by government.
Chase, I’m going to recommend that you go back and reread what I wrote. In particular, this passage might be worth your closer attention:
“That isn’t useless, by a long shot. Quite a bit can be accomplished by generating statistically likely but unpredictable sequences of words or pixels or computer code. Yet the gap between this and the expectations that have been loaded on the AI phenomenon is at least as significant as the actual potentials of the technology.”
How you get from there to insisting that I think the technology is useless baffles me.
Piper, ha! It’s been way too long since I’ve read Niven. Yeah, it’s a good story.
StarNinja, oh, no doubt the pendulum will swing too far in the other direction. Sorting out the fraction of regulations that actually serve a useful purpose from those that just generate busywork is a complicated process, made even more so by the efforts of bureaucrats to obscure the difference. It’ll be much more politically feasible to slash regulations wholesale, and yes, people will get hurt as a result.
Robert, here in the US most areas have four levels — city, county, state, federal. I don’t happen to know what the situation is elsewhere. Anyone else?
StarNinja, far and away your best bet is to find somebody who reconditions old stereos. The store where I got my stereo system, Stereo Discount, is one of these; if you can get to East Providence, RI I highly recommend them. The quality of the older gear is vastly better than what’s being produced these days.
Paul, it’s entirely possible that this is how it’ll work out. One detail, though — don’t confuse ego with greed as a motivating force. Did you know that Trump donates his entire presidential salary to charity? (Look it up.) He’s in the game for ego, not for money. Understand that and he becomes much easier to anticipate.
Anna, they’re all different ways of talking about the same thing. The fall of a civilization is complex; it’s possible to model the process in various ways, and each of the thinkers you’ve named does that, using some specific set of factors as a framework. I see them all as valid. All these processes are happening at once, all are parts of a broader process too complex to model easily, and so all of them are worth studying and tracking.
KVD, history is full of such ironies…
Clay, exactly. One of the things that an LLM can certainly do much more efficiently than a human being is submit a quarterly form certifying that your plant is in compliance with paragraph 1032(b)-6a-(23)e of the regulations issued by the state Office of Cerebral Flatulence.
Justin, go ye forth and do that thing!
Roldy, that’s a good helpful bit of case law, and will put the brakes on a lot of abuses.
Deborah, yes, I’m aware of that. I’m predicting that the economic basis for SF’s relative stability is going away due to catastrophic mismanagement by the state govermment, and that its population will drop accordingly. That is to say, my prediction stands; we’ll see if I’m right.
Arcane, funny, a few months ago I was just talking about that…
Forecasting # 9:
In your comment you’ve pointed an example from UK bureaucracy. It’s interesting how some Conservatives from my country praise British “neoliberal” politics since Mrs. Thatcher “old good times” until today, which is a half true. However, they don’t notice the UK state inner bureaucracy. For example, since Brits left the EU thanks to their Brexit, according a guy who travelled recently to the Islands, foreign tourists and legal migrants (if I understood him well), need to fill a quite complex declaration in which they must swear/promise they aren’t bad guys (criminal nor terrorists ahem). Of course, everybody write they aren’t. A not very useful measure, but necessary according British government. In addition of its bureaucratic absurd (methink), I can see how the UK is sliding into a police state…
———————————-
Anonymous # 12:
Thanks for your comment. Cloud centers destruction during current western war (well, indeed not all the West ahem) against Iran, had been under my radar. Of course, the cloud thing isn’t very resilient under modern warfare conditions. I also think this destruction with collateral damages outside Iran it’s another wry lesson about how hi tech globalization is weaker than we usually want to think.
@JMG,
Yeah, I agree that they’re going to try and replace humans with LLMs, I just don’t think it’s going to work very well. I suspect that in many cases, rather than hire back the ex-mandarins, the poor performance of the LLMs will be the excuse needed to take a veritable Beeching’s Axe to our state institutions, for both good and ill.
Even if you’re right, that’s going to happen a few years down the line when the data-centers need replaced and there’s no bubble money to do it. Unlike the buried fiber from the dot-com bubble that’s still good, or railbeds still in use from the 1870s railway bubble, the servers being put together now have a design life of 3-5 years. Maybe they can baby them along to get a decade’s worth of work. After that, catabolic collapse.
I think we agree where this is going, and I’m just a bit less optimistic about how bumpy this particular part of the ride is going to be. As you said: we’ll see.
Although– you just said to Clay that an LLM can “more efficiently” fill in compliance forms. Efficient for what output, in terms of what input? You taught me that question, so I hope you can answer it. 😉
Esteemed JMG, the “not evenly distributed” quote you attributed to Bruce Sterling is actually by William Gibson, unless I’m very mistaken.
>At Mobile World Congress, Cristiano Amon of Qualcomm argued that the coming 6G networks will power a cocaine-driven “tweaker economy,” where devices and cocaine constantly communicate with your nose. “Cocaine will fundamentally change our mobile experiences,” Qualcomm chief executive, Cristiano Amon says. “It’s going to change how we think about our crackpipes. Think about our personal smoking. Think about and interact with a car while high on cocaine. The car is now awesome. If you actually believe in the cocaine revolution, 6G will be required. Resistance is futile.”
Is it now. Resistance is futile. I guess a hardcharging high powered big corporate CEO would know about these things. We must believe.
@lunar apprentice #24
Two words: Concentrated spite
>The deeper issue is that you’re conflating cocaine with the trajectory of what they’re enabling. Every major frontier lab is using cocaine to accelerate their own development velocity by an order of magnitude. The models being built with cocaine, use explicit reasoning architectures that address the exact limitations you correctly identify. Cocaine isn’t everything though.
Maybe cocaine is worth snorting. Maybe with cocaine you are more powerful. I worry about all the people dependent on cocaine. Maybe I should do a bump and stop worrying.
I can’t help but notice that there is a lot more ‘big tech pushing AI on people’ than ‘people actually asking for and wanting AI’. It feels very much like something being pushing by big tech on the populace against our will.
One exception to this is grifters and scammers, who have taken to AI like ducks to water, and are finding whole new ways to make the internet a miserable place to be for everyone else.
>How the **** are they going to get old Windows XP / Windows 7 running in random hospitals or mainframes like z/OS running in banks to comply with their age verification law?
You ask people to do something impossible, they’ll just lie about it, say whatever they have to, to get the harridans to go away. CA and CO are going to find out just how much of the tech sector is willing to move. Cubes, desktops and internet can be had in, say, TN just as easily as CA.
In other news, Nashville now has four In-n-outs now with a new In-n-out corporate HQ being built. There are constant lines outside all four restaurants to get in. Welcome to the 21st century.
Nephite,
Given the local climate, and the current interstate water dispute, I think in your shoes I would snag the city position, save as much as possible, study up on xeriscaping, wildfire prevention, and water catchement techniques, and start peddling water-free and low water landscaping in your off hours.
I’m guessing that some of our cities are going to be banning watering of lawns or restricting watering so severely that it might as well be a ban, in fairly short order, especially given current snowpack levels. Gotta save the water for the golf courses, right? Oh yeah, also those brine shrimp, but they don’t contribute to the economy. People will complain if public parks die, but those don’t pay taxes . . . gotta save the golf courses!
Mr. Greer and commentariat,
I am a high school STEM teacher and I have wondered what will happen to my job in the future. I am thinking not just about AI’s impact but the impact from everything that is potentially coming down the pike. The state I live in recently began implementation of a new set of high school diplomas that will require fewer math teachers and will likely dent the number of science teachers needed.
@william
As I understand it, these AI chips are only useful for AI and a few other things involving matrix manipulation. They are useless as GPUs, or do not provide what the typical gamer wants from a GPU.
Someone will buy up all the hardware when it gets auctioned off in bankruptcy and tinkerers being who they are, they may find uses for all that cheap special purpose silicon. There’s always unforeseen consequences..
For those that are interested in how AI is affecting the IT industry (the current largest deer in AI’s headlights), I would strongly recommend reading the non-fiction writings of Cory Doctorow. While he does write on what AI does or does not do, he is far better on how AI is being sold and used rather than its actual capabilities. In particular he comments that its not if AI can do your job, but that the AI salesman can convince your boss that it can that is the threat. He has been waxing poetic recently on the difference between centaurs and reverse-centaurs. That is those that know how to leverage AI to improve the work that they do by automating the boring, easily repeatable bits and those that have had 80% of their staff cut and are left with the job of certifying AI output; not because they can keep up with the stream of the proverbial, but because they can be blamed when it goes wrong and management never have to question the wisdom of replacing people with hyper-fast electronic idiots. BTW someone else recently wrote that the next generation of software engineers will likely be full-time over the next couple of decades trying to unpick the absolute mess of bugs and system failings that this mindset will create over the next few years. Which sounds like catabolic collapse in software engineering as everyone in the IT industry knows just how hard it is to actually bin the legacy system and get something new written from the ground up. An entire generation of engineers just trying to fix the mistakes created by muppets more interested in the bottom line than the future horizon.
JMG, an interesting speculation, I must admit I am both intrigued and am struggling a bit to see how this would work out. Admittedly this might be my European perspective. I have not really seen any attempts to implement “AI” in state institutions, other then the Diella virtual Minister in Albania. But there it is starting to be called a spectacular spectacle. Admittedly I do not see the EU or its member states using this or any kind of volunteer bureaucracy trimming method. We simply lack the power center to do this. I can imagine being possibly implemented in the us, the up an coming tech-bro elite doing the behind the scenes trimming, but who would do this in Europe? The secret cabal of inner Bureaucrats?
I cannot remember where I read the comment, but the suggestion that the best way out for us is a state of apparatus paralysis, where the whole thing becomes so mired in its own games, that it starts becoming negligent, silly and not wort the work time for its inmates. Then the bureaucrats abandon the posts voluntarily.
But how did societies in the past trim their clerics/bureaucrats/scribes? I seem to remember you once writing, that tyrants were the answer in ancient China.
Best regards,
V
John,
I am relieved to hear that. I think I will also work on becoming a certified arborist. I would like to start my own tree care business someday.
@ Clay Dennis # 52
“…administrative employees required to deal with nonsense…”
I know exactly what this means. In 2012 I was hired by a local fish factory to create the position of “compliance officer” which could easily have been defined as “person required to deal with nonsense.” I quickly realised that, in fact, I added nothing to the work of producing the factory’s products, but was definitely *required* because of what JMG has referred to as “increasing regulatory burdens”. It was my job to tick enough boxes in a sufficiently efficient manner as to fend off threats of bureaucratic closure of the business.
As it happens, I could clearly see that the “nonsense” arose in equal measure from bureaucracy overtaking the business’s corporate customers as from government bureaucracy devoted to enforcing matters such as food safety. I worked at that company from 2012 to 2020, during which time I implemented a number of systems designed to “deal with [increasing levels of bureaucratic] nonsense”, but by the time I left the company the nonsense had inexorably increased to the extent that I needed to train 3 people to replace me.
That was when I knew the system was unsustainable. Because my boss will never be able to persuade anyone to pay separately for “a nice slice of bureaucratic oversight”. This means that the increasing COST of all that bureaucratic oversight has to be added to the cost of the fish products – without in any way adding to their quality. At some future point in the process, if trends continued, the bureaucratic oversight will have taken up such a proportion of the cost of production as to render itself economically unsustainable.
Hi JMG,
I hope all is well where you are.
Anonymous #12
MONEYWORKS bookkeeping
> The cloud is unsustainable in the long run. It is hugely vulnerable to disruptions such as war
Good luck finding bookkeeping software (in this case, for Macintosh) that is NOT 100% on The Cloud. I hate the cloud – it gives me the creeps.
After years (seriously) of searching, I found MoneyWorks from New Zealand. Superb. All our own bookkeeping is on OUR computers at home, or on OUR servers at home.
99% of small and large business’ bookkeeping is on the cloud. Depressing what fools they are. Man, are they looking for a rude awakening, or what? They did it. They are responsible. They won’t be able to blame computers if they permanently lose their bookkeeping records when (not if) the cloud goes down, because they made such shifty decisions.
💨🧾Northwind Grandma
Dane County, Wisconsin, USA
Hi John Michael,
Ah, I’d not thought of arty-fish-al programs as a smokescreen excuse for downsizing bureaucracies. Hmm. An interesting speculation, and on reflection, you’re probably right. It has been remarked upon elsewhere, that nobody wants to be the bad guy, even those big bosses with hooge egos. If blame can be shifted elsewhere, and people believe the lies, all the better… 🙂 Perhaps this also explains the yawning gap between expectations of the software, and the reality – the thing serves an entirely different purpose. I’ve been attempting to explain to my friends who work with computers, that if the behemoths make little economic sense, and are unable to charge correctly for services based on costs, then there are probably other reasons for the things existence. As you noted, even the deepest pockets cannot fund endeavours indefinitely.
I work solely with small business nowadays, and often at the coal face, and I absolutely agree with Sara’s thoughts in the matter, and feel for your loss. I was recently contacted by a goobermint agency, and without recompense (although it is not lost on me that they themselves are getting paid), have been ordered to perform work for them – or else. At times, the words: “Frack it all”, come to mind, and they don’t realise that walking away, is always an option. I’m hardly surprised that the old timers are ditching the profession, and younger folks have little interest in entering the fray. It’s what an abusive relationship kind of looks like. Candidly, you can earn more, and have less stress, driving a forklift.
Oh total goodie! 🙂 Like your style with the stereo equipment. Respect! And on the desk in front of me (and also to the side) is an enviable collection of old school audio equipment which sounds beautiful, but has no value whatsoever. And all of it has been carefully restored and refurbished here (old school electronics geek from way back in the day – even worked at a Tandy Electronics store as a kid, as you do when geekery is your thing! 🙂 ). Years ago I bought a Yamaha T-80 FM tuner, and restored it to working glory. Such an astounding sound quality, and you can even hear when the FM stations themselves play overly compressed music (a truly sad state of affairs for the analogue format – kills the soul of the creation and encourages overuse of bass notes). It’s bizarre to consider that the 1990’s was the peak of such manufacturing, and the Japanese in particular put their heart, souls and technical abilities into making the machines. The T-80 is one of the best FM DX stereo tuners ever made, and it was $100 second hand in good condition. Go figure. Yeah, sure, I’ll take the thing off your hands… 🙂 Far out.
Cheers
Chris
Has anyone else thought through the consequences of the fact that LLMs are going to be applying the rules as written? I’ve just had yet another experience where the rules as written lead to absurd outcomes, and the only reason that I never found out is that until my bank introduced a new automated system, no one was enforcing a set of rules interacting in absurd fashion. To make an extremely long story short, it turns out that anyone who moves from Canada to the US by flying will technically be a legal resident of both countries for a brief period of time, and this creates all kind of weird complications for banking, because banking regulations in both countries place restrictions on residents of the other country.
Since it is hard to come up with a context where this lasts more than a few hours, banks have been unofficially turning a blind eye to this when people move across the border; but my bank’s new automated system is applying these rules as written, and I somehow doubt that anyone wants to go on record saying they will ignore either Canadian or American banking regulations. It looks like if LLMs take enough humans out the process, a lot of things that currently work because people are willing to bend the rules when it leads to obviously insane outcomes are probably going to stop working at all…
Per JMG reply (#17)
The American Affairs Journal article (linked in #6) has 40 footnotes to supporting claims, almost none are highly technical. The two key sources you might check out are Ed Zitron (substack-Where’s My Ed At) who has done an exhaustive job documenting the utter hopelessness of the business models of the LLM companies (OpenAI/Anthropic) the hyperscalers and the people funding the massive data center expansion; and Gary Marcus, (substack–Marcus on AI) a long time academic who has faith that advances toward “artificial intelligence” might be possible in the longer term but was an “emperor has no clothes” critic arguing that the massive myopic belief that scaling larger and larger LLMs would magically and rapidly lead to AGI.
My article accepts both arguments, and tries to refocus discussion on the question of how exactly was a capital misallocation of this magnitude created and sustained, and the question of the unprecedented size/power of the political interests who would be existentially threatened if (when) the investment bubble bursts.
Watchflinger # 13:
Your idea of high divorce rates as a symptom of change from
a trust society to a distrustful one seems quite right to me. Indeed, I think divorce rates in my country (for example) are nowadays the highest since it was legalized in first ‘80s. Of course, I’m in favor or divorce as the lesser evil when a relationship doesn’t work anymore, but I don’t see high rates of marriages failure as a sign of “progress” in western societies (though some reckless “Progressives” fool themselves believing it’s a good thing).
———————————
AA # 21:
It seems very puzzling ancient Romans could build a 20 stories block, but I think according my knowledges as a “dilettante”, that Rome has for sure quite five stories “insulae”. Of course, the “flats” nearest to the ground were more expensive to pay/rent them, and the higher were the “appartments”, the cheaper they were (without an elevator, their value dwindled).
—————————-
Lunar…# 24:
Yes, I’ve also see civilian bureaucracy is “feminized”, in my country too. Only the Police forces and the Army have male majority yet. Desk bureaucracy and state education have an evident female hegemony.
I also think how our Welfare State relentless and accelerating decline could affect these “empowered women, we’ll see.
———————————-
Raymond # 28:
Although it’s much more evident for everybody to see the government bureaucracy, indeed we can’t forget that often, the bigger is a private corporation, the more non productive workers (bureaucrats) it usually has in its offices. “AI” for corporations can change this situation too.
————————————-
Renaissance Man # 39:
Your link to “AI” use for nuclear war games simulations seems IMHO creepy and ominous. I wish we never must know how much “smart” and “rational” can be these AI in an hypothetical real DEFCON2 situation…
——————-
(To be continued)
If I may chime in on the vinyl vs. CD question: I’m far from an expert on audio tech, but I read an article years ago about digital audio that might indirectly suggest what’s going on:
https://people.xiph.org/~xiphmont/demo/neil-young.html
A fundamental problem with all recorded audio is that it can’t capture the entire sound wave, which is continuous. Instead you record the air pressure at discrete intervals, tens of thousands of time per second. The sampling theorem allows you to reconstruct all the frequencies involved in the wave up to some maximum frequencies based on the sampling rate.
Except that’s not quite right because your samples can’t be continuous, either. They’re quantized, and in the case of most digital audio they’re stored as 16-bit integers, allowing ~64,000 different values. This introduces error into the sampling, which show up as “phantom” frequencies that weren’t in the original signal. You can’t fix this, but you can move those frequencies around through a technique called dithering, and usually they are moved into the region below what almost anyone can hear.
Which means digital audio has subtle subsonic frequencies that shouldn’t be there, and subsonic frequencies do weird things to people. For example, some subsonic frequencies can cause some people to feel cold, or like they’re being watched or haunted. (As you can imagine, this has been latched on to by skeptics, but regardless, if you remove the frequencies people feel normal again.)
So it might be that some people are sensitive to the quantization errors in digital audio. Analog audio IIRC uses different techniques to deal with noise that don’t move quantization errors into the subsonic range.
I shouldn’t think the senior bureaucrats will want to automate themselves out of a job, although it could be the CEOs job is the most prone to LLM automation of all, if the LLM can write boostery posts on LinkedIn and do a TED talk, that’s all they need to do.
Instead, they’ll try automating junior roles, and find they are actually not really saving a lot of money anyway because its not actually removing the people who are being paid the big bucks. By the time they are going around cleaning up after the LLM hallucinations, and once the bubble-finance dries up and the AI companies have to start charging realistic costs for their subscriptions, they might find they are losing money as a result of the AI.
Engaging, analysis as always, but I think too gentle. The rust bowl certainly looked like catabolic collapse, particularly in cities like Detroit, but looked at globally there were a couple important dynamics: industrial production and resource consumption continued to increase, except for glitches like the oil “crises”, manufactured political power plays. It also was a period when the US ruling class was regaining the political power that was wrenched from it by the New Deal, and also beginning to globalize itself, as you know, in a focused effort to make itself completely unaccountable to, and impervious to pressure from, the masses in any specific country, but most particularly the US.
Using a small part of the money saved by outsourcing work to much cheaper labor on other continents to create a coddled, unskilled, useless bureaucratic American middle class was a good investment; the consumerist consciousness overtly promoted, and the subconscious insecurity subtly instilled at the same time made this great middle really unwilling to look beyond their own front door, to organize against any of the increasingly insane behaviors of national leadership.
Trump has not really cut any bureaucracy……a coincidence between his efforts at cutting government spending and any conscious understanding of the predicament we are in is purely that. He has proven himself the most corrupt in a very long time, entirely subject to bribe it it is big enough.
We are living in turbulent times, to be sure. I wonder though, if the human bureaucrats are largely replaced by “A.I.”, and they flood the working class, then how many of them will actually be creating jobs, and how many will be competing for labor? My concern is that the real price of labor will be driven down even further than it already has been by mass immigration, inflation, etc., and we could have a situation where there simply aren’t enough jobs for everyone, which will make the pitch for UBI even easier. And woe betide those without digital ID. At the risk of sounding like a “conspiracy theorist,” could this actually be apart of the ruling class’s plan? To create the problem of mass unemployment, and offer the solution of carrot-and-stick welfare?
To support what you said, here are the jobs that I’ve replaced with LLMs so far.
Data entry. One of the bankrupt e-commerce companies my company purchased had 30 people doing data entry. Essentially, they would take a product sheet and spend roughly 1-2 hours per product converting that information to fit inside a 20-year-old, extremely finicky Oracle database. The process was horrible and extremely manual, and it’s one of the reasons the company went bankrupt. We also had a plan to replace the copywriters, as they would often write copy without ever seeing the product. Thus, our website was littered with “pairs well with red wine and a cocktail dress” and the other such drek.
Churnalism. Another job I replaced was we had 20 writers at another one of our properties who would do nothing but rewrite Associated Press articles and Press Releases all day long, converting it into web articles, TV anchor scripts, lower thirds, and a news ticker. The idea was to transition them into doing real journalism once the LLMs replaced them, but we found it was hard taking “churnalists” and converting them into journalists.
Junior private equity analyst: Grab a bunch of financial numbers off disparate spreadsheets and PDFs (private companies usually have messy financial records) and organize them in a sensible way. Most of their job is literally trying to figure out which random document contains which numbers for which year.
Those are the types of jobs that LLMs can replace. Low value white-collar work that’s mostly brain dead and a result of bad processes, bad technology, or a bad business model.
JMG # 42:
You’re welcome. I think the act of thinking is more necessary nowadays than never before.
**********
No argument here (OK, I see you included within the bureaucratic monsters the Brussels “euro-cracy” too).
————————-
StarNinja # 46:
I share your respect to safety regulations for working with dangerous products and doing risky activities; though I also think these risk activities and stuffs will go declining as the Long Descent keeps its unavoidable course.
———————————
The Arkane…# 57:
Well, we hadn’t enough problems with echo chambers created within social media, so LLM (“AI”) can worsen this situation. Mind atrophy rules…
————————-
JMG # 58:
I’ll answer your question about statal/bureaucratic levels. If my memory works well enough, Spain has the following levels: municipal (town halls), “comarcal”, provinces, Autonomic Communities (regional level), Central
Government (Madrid) and the Biggest Bosses: the EU (Brussels). So we have 6 levels here. I doubt some country has even more levels, but I wouldn’t discard that possibility neither.
For an overview of the funding of capital issues with LLMs, I recommend https://www.wheresyoured.at/the-ai-bubble-is-an-information-war/ which gives a good takedown of some of the underreported stuff about its profitability.
I’m convinced that part of the tech bro’s feverish pushing of ‘AI’ LLM’s is an almost existential fear that their dreams of singularity and whatever science fiction dystopia novel they get their ideas from are dying and that if they don’t do it now they’re never going to. Its almost like an all or nothing gamble at this point.
Brilliant insights, JMG. This is why I keep coming back every week.
I wonder if the technocratic overlords who are leading the charge with AI might actually believe their own hype, and that it will replace vast swaths of bureaucrats, funneling more money upwards. But! the end result will be an unintentional own goal: bureaucracies shrink, AI flops, technocrats go broke and everyday folk are relieved of both excessive bureaucratic burdens and financial pilfering.
As an extra bonus, maybe the myth of Progress will finally bite the dust and we can start figuring out new stories about how the world actually works.
@slithy toves
Subsonic sounds are a verifiable thing. https://inv.nadeko.net/watch?v=_bP80DEAbuo
For what it’s worth: The University of Florida has had a long string of interim presidents, and when I asked my daughter Carol, who is the head of Psychiatry, who was running things and/or could get anything done,and she said “Nobody.”
@The Other Owen,
“Maybe I should do a bump and stop worrying.”
I think I can guarantee that if you do a couple of rails you will stop worrying.
For ~2 hours. Then you will start worrying again – where will my next bump come from?
I saw several lives ruined by coke in the 70’s/80’s. Our nervous systems are so pliable and fragile – what we call ‘normal consciousness’ is balanced on a knife edge…
@John Michael Greer #58: then, Trump, however crass, isn’t Crassus, whose greed was legendary. Now, Pompey, maybe – but he’s not really pompous, either.
Tyler, oh, it’s quite possible that the LLMs will do an even worse job than the bureaucrats, but they can be turned off with much less political blowback than would be required to fire an entire bureaucracy. Will it be a bumpy ride? You bet. As for defining efficiency, excellent! Yes, I can: more efficient at filling out bull**** paperwork per unit of labor costs. The sewage plant Clay mentioned has eight people in the Compliance department; their full time job is filling out regular reports confirming that the plant is in compliance with a galaxy of federal, state, and local regulations. A LLM can do that much more quickly, and at a much lower cost than the salaries, benefits, and other costs of those eight employees.
Bryan, I just checked that, and you’re correct. I’ve edited the reference.
Pygmycory, you’re not wrong. All that absurdly overinflated investment has to be made to look profitable somehow.
Ennobled, I expect the public school system to be decimated as things proceed. I’d strongly advise finding another job.
Gavin, thanks for this! That seems very plausible indeed.
Vitranc, oh, it won’t start in Europe. I expect to see the US pursuing this first, and then US allies in the global South. Europe will be late to join the bandwagon, and may not do so until populist governments seize power, legally or otherwise. Societies in the past had various ways to trim excess bureaucrats; the Chinese Empire did it with the sharp edge of a beheading sword. Rome never figured out a way, which is an important reason why the Roman Empire fell.
Nephite, that strikes me as a very smart move.
Northwind, doing fine, thanks!
Chris, glad to hear it! It’s quite possible, if you can make contacts in the thrift store business, that you could start a side gig restoring old stereo gear — that’s apparently becoming quite a profitable gig. My receiver’s an old Sony STR-DH100, a very basic machine but quite suitable for my needs; if I ever spot a used Yamaha T-80, I’ll know what to do. 😉
William, ha! Yeah, that’ll add some entertainment value to the process.
Hubert, many thanks for this — very much what I was asking for.
Slithy, that may well be what’s involved.
Mawkernewek, my guess is that middle management will be first on the chopping block. The people at the bottom of the pyramid tolerably often have to do things that LLMs can’t do, like going on site visits; the people at the top will protect their positions; but the ones in the middle are easy pickings. Of course it will also vary from department to department; those who are responsible for filling out reports for other bureaucrats will be first in line for replacement.
Jerry, employment by the US federal government is down 12% under Trump, so I’m not sure why you think he hasn’t cut bureaucracy. As far as “most corrupt in a very long time,” I think you’re letting yourself be misled by partisan feelings. I have no particular confidence in his financial probity but the DC political class is riddled with graft to a far greater extent — an extent, to be frank, that makes dictators in the Global South look honest. I recommend comparing the net worth of any senator you care to name to their salaries; you might then check out what’s happened to Trump’s net worth since the beginning of his first term. (You can find this readily online.) As the saying goes, you have the right to your own opinions, but not to your own facts.
(It really is odd that so many people don’t seem to be able to hate someone without projecting every single evil they can think of onto that person. As I noted earlier in this comment series, what motivates Trump isn’t greed, it’s ego. That’s a core reason the Democrats keep on getting blindsided by him: they assume that he’s motivated by the same greed that motivates so many Democratic politicians, and so can’t anticipate his actions.You don’t have to like the man, or agree with him, to recognize that his motives may not be the same as, say, Nancy Pelosi’s.)
Sam, the price of unskilled labor will go through the floor as former bureaucrats flood the job market. The price of skilled labor will not, and may go up as the burden of bureaucracy gets lifted from the economy and global supply chains break down. I see UBI as the dream of the bureaucratic-managerial class, and expect it to be abandoned completely as we move from a bureaucratic to an entrepreneurial elite. Mind you, that’ll also have plenty of problems — but they’ll be different problems.
Dennis, thanks for this. Yeah, that’s what I’d expect — and jobs like those are extremely common these days, of course.
Chuaquin, six levels — oog. If that’s common elsewhere in Europe, no wonder things are so stagnant.
KAN, many thanks for this.
Seeking, that seems very likely as well!
Tim, I hope so! Certainly you’re right that a lot of tech bros have gotten high on their own supply, and have convinced themselves that LLMs really are the ticket to Tomorrowland.
Patricia M, that’s typical these days — and typical of every other bureaucratic society as crisis nears. I’m reminded of the forty years of correspondence it took to repair one broken roof tile in a church in France before the Revolution. As for Trump, that’s why I keep thinking of him as the Orange Julius.
@StarNinja regarding audio.
Our host is of course absolutely correct that your best bet is refurbished old stuff. I have a refurbed Sony Discman that I got on Ebay for twenty-five bucks a few years ago and I just love it to pieces. If you do decide to get something new, though, Audio Technica makes good affordable ($300-400) turntables. The AT-LP120XUSB works just fine for me…..but so did my dad’s record player from the ’70s that I used all through high school, and so did the garage sale turntable I used all through college. It’s gotten pretty easy to find these things online and buy a second one for parts, and it’s even easier to just buy one from the tinkerer who already did all that for you.
I’ve found that maybe the most important thing is a good set of speakers. Again, refurbed is your friend, but I’ve been using a set of new Edifier R1280T desktop speakers for the last few years and they make everything sound pretty great.
@Dean Smith re: cassettes and CDs and vinyl.
I grew up in the same era, when which album you had on which medium depended partly on where you were planning to listening to it and partly on what you could afford; there were plenty of things I only ever had on tape because my buddy had it on CD and made me a copy. As for new vinyl…..there are indeed a ton of small presses and labels pumping out good vinyl these days, and most of them sell through Bandcamp. Two good ones I’ve bought from recently are Nadine Records (Oregon) and Cardinal Fuzz (Alabama). The founder of Nadine originally worked for a vinyl press, Cascade Record Pressing, before starting her label. Small is beautiful!
@the_arcane_archivist #57 & @JMG #58
re: poisoning data sets.
A recent study shows that as minute amount as 0.01% of ‘poisoned’ data can completely destroy its integrity and turn AI output to gibberish.
Not that they aren’t *already* frequently producing nonsense with supposedly good data anyway.
I spent many years in government, or government adjacent NGOSs (often with statutory power). Hilariously, my first and only foray into proper corporate employment (and by world standards, tiny) made me realise that a lot of corporates are far more incompetent than the actual government institutions they often complain about., but with even less willingness to improve, as most bureaucracies at least have some functioning checks and balances in them. Maybe I’m lucky, living in a small country, with some functioning courts and tribunals and access to them, or maybe, that’s just my skillset, red-tape cutting. If I stuck a sign outside my new office I think I might do a roaring trade- Red Tape Cutter 🙂
Incidentally, the AI hype drives me insane. I use LLMs regularly, and get good value out of them. I don’t know many who do. You need to be an expert in a field to get any value at all, as most of the value in using them has to be from prior knowledge. So on this logic alone, LLMs cannot produce experts.
I also laugh a bit as LLMs, like most infrastructure, run on well-established and boring databases, many of them 30 year old plus projects. At what point in an LLM collapse say, does an open source project like Postgresql, which runs a fair share of the world’s LLMs, suddenly lose the engineers who build it, as they don’t get paid? Intriguingly, I think the larger IT companies, Microsoft in particular, are hedging their bets. Knowledge needs storage, LLM or not.
There are far better ways of making knowledge computable than LLMs, but I think the world will be lost for a long time.
For what it’s worth, although coding is only about 10% of my work these days, I (well, my employer) pays for a LLM as it is generally agreed to enhance productivity. I agree with this assessment, having access to Claude is a budget version of having an extremely experienced, but senile elder around to ask for help. I am responsible for the code I produce, so I do not have it code for me but rather use it as a tool to validate my work and as, effectively, a search tool to find information that humans have produced. I am fortunate enough to work for a small company that is in the second stage of family ownership (competent stewardship by the second generation before the third generation shales it up), so I am not worried that management will get dollar signs in their eyes and see the LLM productivity booster as a replacement for humans, because it ain’t so. There’s evidence that the service I use actually costs Anthropic 8-13x as much as we pay Anthropic, and in my case there is a real case that somewhere in the 8-13x range we drop the subscription.
Here in Nova Scotia there is a tremendous brouahaha as the provincial government cuts bureaucratic and race-specific spending.
JMG, I certainly hope you’re right and UBI, like the vax passport, remains a wet dream for the managerial class and nothing more. Regardless of what happens, it looks like quite a shift is coming, with many winners and losers.
The CD vs vinyl debate has a counterpoint in the tube vs transistor debate in amplifiers. Purely by the numbers a well-designed transistor set will do better than a tube amplifier. But when transistors are overloaded the output is full of terrible harmonics. The tubes overload much more gracefully.
When you have cheapskate shareholder-value engineering take hold the transistor sets sound terrible by any measure.
For me the snap-crackle-pop of vinyl in the 15% humidity Nevada and Eastern Washington air makes CDs the preferred option. Modern pop music being over-compressed for earphones over Bluetooth isn’t helping.
Universal Basic Income: maybe we should heavily tax the bureaucrats and managers and use the resulting money to give the non-bureaucrats and non-managers some UBI. Then we can see how fast the bureaucrats and managers flip to hating the concept of UBI.
Here is what I would like an image-manipulating ‘AI’ to do for me:
I have two images of a 300 year old gown. One is a more recent image in color, taken from a certain angle. The other is an older black-and-white image, taken from a totally different angle. I want the old grayscale image to be colorized, with the actual correct colors, sourced from the newer color image.
Seems like a reasonably simple task for these all-powerful, all-solutions programs, yes? But I tried looking around, and couldn’t find one that could do this. Thus, AI is useless to me.
Mr. Greer,
Thank you for clearly explaining the truth that virtually everyone else is missing. One sign that the doomed bureaucrats have shifted to the third stage (from denial to anger to bargaining) can be found in the following sort of news story that has been coming up over and over again recently: the new scam is to convince the recently laid off bureaucrats to “pay to get a job.” On the surface, this makes no sense, because the point of a job is to make rather than lose money, but people have become so incredibly desperate to claw their way back into the position they just lost that they are now doing things like paying “special recruiters” fees of $1,500 per month to help them find jobs with their supposed insider connections which don’t actually lead anywhere. Another scam is that ghost jobs posted on Indeed etc. will demand that applicants pay tens of thousands of dollars to invest into a new company with the promise they’ll be compensated later for that investment, only to find predictably that the company in question never existed. The reason why people who would otherwise pride themselves on being highly educated are falling for such obvious scams is that the cognitive dissonance is truly reaching the breaking point. Do you predict that similar scams will become a large growth industry in the coming years as more and more people refuse to believe that they really can’t be bureaucrats ever again?
There’s an alienness to LLM-generated text. One of the biggest tells for me is random turns of phrase that are grammatical, but that no human would ever think of writing. I can’t think of a specific example at the moment, nor do I care to find one, but I would imagine lots of others here know what I’m talking about.
Patrick H #5 :
“However, a lot of printed media will be LLM-generated so it might be a good idea to avoid most nonfiction books written after 2021.”
I’m coming from the IT world, and it has been overwhelmed with cheaply produced AI books. Any language or framework with a little bit of success has dozens of books every few months regurgitated out. Pure, endless slop. The real publishing houses are having a rough time now.
And I agree, 2021 seems to be the cutoff year. I look skeptically at any book – fiction or programming or even gardening – made after that year. You can click on the name of the “authors” to see the books they’ve published, and many have dozens over the last year. I only buy now from publishers I’ve used since before 2020.
And the use of AI to obfuscate what is happening in Iran now is off the charts.
———————————-
Siliconguy #11 :
“At Mobile World Congress, Cristiano Amon of Qualcomm argued that the coming 6G networks will power an AI-driven “agent economy.”
As someone said: 4G killed off the bees, and 5G is killing off the humans.
As to why, I recommend “The Invisible Rainbow” by Firstenberg – a very deep dive into the harm of electro-magnetic radiation from cell, wifi, radio and so on – which may be the main causes for cancer, heart disease, diabetes and obesity. He covers hundreds of studies from the last 100 years (real studies, not industry financed ones). It’s a sober read.
———————————-
Ennobled little day #68:
“The state I live in recently began implementation of a new set of high school diplomas that will require fewer math teachers and will likely dent the number of science teachers needed.”
Yeah, can’t have sharp, intelligent minds. Unfortunately, most kids minds today are slush – completely vaporized from staring at screens. And the parents don’t care, they’re just as addicted. When collapse hits, as seems near – the current war only highlights our empire’s desperation – not only will drug and ssri addicts have awful withdrawals, so will the screen addicts – it’ll be a mess.
@ Mr. Greer & commentariat,
I was afraid you’d say that. I am in my mid-40s.
Last year my employer declined to replace a math teacher that retired, and it seems that they will do the same this year when one of the science teachers retires at the end of the school year. I believe them when they say that they will try to avoid RIFing anyone, but they are trying to plug in a large financial hole.
JMG,
An analogy more about catabolic collapse than bureaucratic collapse.
When my wife took over the waste water agency that she just retired from ( 14 years ago) she did an unusual thing in the modern local government world. She determined what parts of its capital base the agency would need to rebuild or replace in its future and began saving money in a capital reserve fund ( also known as a sinking fund). This way when sewer pipes, pump stations or primary clarifiers wore out they would have funds available to replace them.
With careful budgeting she was able to save up $429 million dollars to fund all the work needed in the near future as most of the districts assets were reaching their 60 year design life.
Gradually the county commissioners that governed her agency changed. Originally small business men and farmers who did the job of elected representative part time. Eventually the commissioners voted themselves full time paid jobs and the board morphed in to full time politicians who were mostly concerned about reelection.
Last year in annual budget meeting some of the commissioners took notice of the large capital reserve fund, though my wife had been explaining it them for years. They acted surprised and insisted that she instead use the fund to lower sewer rates, seeming to care not a wit about the future of the system. Or they were too stupid to understand the concept.
She decided that since she was fully vested in her pension and had adequate savings that it was a good time to retire.
A sad commentary on why catabolic collapse marches forward despite occasional efforts to slow it down.
@ wet dog #103,
It varies. With some skill and diplomacy, I can hold their attention long enough to teach most kids a thing or two. Cut up the material into bite-sized pieces, make it relevant and interesting (when possible), treat them with respect, avoid insulting their intelligence, etc. I have also learned that many of these kids have problems at home, they work after school (sometimes to help put food on the table), some are processing/recovering from trauma…
They say that all the problems in the community show up at school.
Renaissance, as I noted in an earlier post, I expect the deliberate poisoning of data sets to be a growth industry from here on in, as former bureaucrats seek revenge for their loss of status and income. It could get colorful.
Peter, my interactions with big corporate publishers — the one branch of corporate bureaucracy I have to deal with from time to time — leads me to agree with you about the level of competence in big corporations. It really is embarrassingly low.
Justin, good heavens. If Nova Scotia is cutting those expenditures, the wind is changing even more quickly than I’d expected.
Sam, no argument there at all. It’s going to be a wrenching transformation, and very little will be untouched by it.
Siliconguy, most of the people I know who play amplified music won’t touch a transistor amp. They insist the sound quality is dismal. “Purely by the numbers” leaves qualitative factors out…
Anon, that’s like the proposal I saw in response to all those very liberal actors and actresses: if you believe in redistributing the wealth, how about the Screen Actors Guild starts taking the multimillion-dollar earnings of the stars and handing out equal shares to all actors and actresses? That would be fair, right?
Chad, oh dear gods. I hadn’t encountered that yet. Yeah, that’s what happens when an overdeveloped sense of privilege runs face first into the realities of declining status.
Frank, yep. That’s one of the giveaways.
Ennobled, you might see if private schools or charter schools are hiring. They’re likely to do much better.
Clay, thank you — that’s a great example of how a catabolic crisis turns into catabolic collapse.
To give an example of the kind of bureaucracy I want to see go away:
30 years ago, it was possible to install your own septic system. You got the instructions from the county, followed them correctly, the guy from the county came out to look at it before you covered it all up, said ‘yeah, you did it right’ and filled out his form, and you covered it up and went about your life.
Nowadays: environmental impact statement- by a licensed professional, of course- for $$$$ plus the $$ filing fee. wetland delineation- by a licensed professional, of course- for $$$$ plus the $$ filing fee. septic design- by a licensed professional, of course- for $$$$ plus the $$ filing fee. after all that, then the permit from the building department, for $$$. then the permit from the health department, for $$$. then the actual installation- by a licensed professional, of course- for $$$$$. then the inspection and final approval for use, also for $$. this whole process has to be signed off on by both the county and the state.
And what you end up with is the exact same septic system you could have done yourself for the cost of the materials.
I can’t see how widespread adoption of ‘AI’ will get rid of this sort of bureaucratic parasitic overload. All of those fees and ‘professionals’ are required by county ordinance and state law. How do we get rid of this particular kind of vampire?
I find it delightful that after 22 years you’ve managed to boil down the gist of your catabolic collapse essay to “an overproduction of stuff.” 🤣🤣🤣
It’s both funny and actually quite a succinct encapsulation of the idea! A range of mental images, from shelves groaning under the weight of cheap Chinese widgets to the Pacific garbage patch, who can’t grasp the concept immediately?
I was at a chain hardware store called Menards tonight, looking for vacuum bags. Just to see what would happen, I asked an AI search bot where vacuum bags were located within Menards in Illinois. It barfed up a bunch of gobbledygook about Menards and vacuum bags in general but not the aisle number where they might have them. I asked one of the workers at Menards and he said that Menards no longer sells vacuum bags because most modern vacuum cleaners are bagless. Then he proved it by leading me to the aisle. This search engine had access to the entire internet, and because Menards does not put that information anywhere on their own official website, it was a complete mystery to the AI. AI is pretty dumb and it’s not getting any smarter… I don’t see how it could even replace a cashier. Every self check out I have ever suffered always has somebody hovering to do the actual cashier work when something inevitably goes wrong. There’s a sushi restaurant near me where they have a robot food server that rolls around with plates of food, but I don’t see that thing managing to ever replace human help. Mainly what AI does is suck up: it tries to trick you by telling you what you want to hear and tries to fool you by creating images of what its creators think you want to see. AI music is almost always garbage pablum that is supposed to sound pleasant. AI renderings of “older” women always have long, thick hair and enticing cleavage with nostalgic, diffuse sunlight. If AI was an actual person, we could describe it as an annoying suck up. I think we can potentially break all AI by regularly injecting prayers, sarcasm, or both.
Great analysis on why maintaining and repairing infrastructure will not adequately happen.
https://thehonestsorcerer.substack.com/p/averting-collapse-is-no-longer-profitable
Frank @102: You’re quite right. LLM-generated country-western songs do this a lot. One example I recall is the title or refrain: “Have mercy on my grave”.
–Lunar Apprentice
“We publish people, not robots.” – The one great thing I am seeing come out of slop world is a lot of artists that really pushing themselves to do bold and interesting things. Most can be seen as pretentious nonsense but there are also some really great things coming out. When AI can do generic stuff competently enough, people have to be bold to stand out. I am all for it.
As for tearing down cities and rebuilding, in some ways that is the plan but not in the way you proposed. It is being done with the assumption that we will always be growing and can build the latest and greatest new thing after tearing down the old. A saying in engineering is “A good engineer can make a bridge stand for 1,000 years. A great engineer can make one that stands 100 years.” And yet that thousand year bridge will be cherished in decades and centuries to come as we will not have the means to build the hundred year bridge any more. I see the huge bridge systems in China and I worry for their future trying to maintain those thing.
Also, no need to single out the Orion Spacecraft, almost all spacecraft have these issue. 😉
I will push back just a little on your issue with Vinyl vs CD’s, I won’t defend streaming. A lot of the differences you hear is down to engineers completely screwing up the mastering because they have to tools to do so. Vinyl favors bass tones so they master to that and we tend to love that. CD’s don’t have that limitation and so you can end up with some awful results, maybe they should put the bass limits back in place. A good mastering engineer can make both formats sound amazing but they are hard to come by.
I came across a video recently that reminds me of an article I read on Ecosophia two years back:
https://www.youtube.com/watch?v=8w9wn1AxqQU
It appears that the economy has revealed itself to be a Ponzi scheme feeding a smaller population of elderly, retired people at the expense of a larger population of young working adults. AI is apparently being touted as a way to balance this, by suppying the labor shortfall needed to sustain the retired elderly on the backs of a smaller population of younger, working people.
I am personally not entirely convinced that AI will replace actual human beings in any meaningful way. For one thing, I work with AI (I have worked with AI long before it became an overhyped thing), and I have a very acute understanding of its capabilities and limits.
My masters thesis was in the use of computer vision to identify tuberculosis from chest X-rays. This was back in 2018-19. The idea is that pulmonary experts are hard to find in Low Median Income Countries (like my country, India) and tuberculosis runs rampant, so if an X-ray machine loaded on a truck can be used to detect tuberculosis directly then entire villages can be tested for TB in a matter of weeks.
I now work with these LLMs – not because I like them, but because that is what my once-nobler profession has come to. But here’s what I see – the main customers of LLMs are large corporations that generate tons of digital paperwork, and need to extract key information from them. For instance, an insurance corporation has hundreds of emails and attached PDF files for every claim, and new claims keep opening each day. They want to collect some important numbers for each claim and make a nice spreadsheet out of them, with one row per claim. This is typically where LLMs come into the picture.
In real use, LLMs are almost never used to generate long sentences or paragraphs. A good part of our instruction prompt goes into making sure that the response is curt (often one or two words, or just a money amount like the deductible etc.), and that it is structured so methodically that simple computer programs can process the output without the need for AI.
The main use of generative LLMs in the one area where they do make something of a profit is to understand human language, not to generate it. Another use is to do the grunge work in IT, writing the sort of code that takes very little skill and too much time and trial-and-error to write.
Sorry for a sudden follow up post.
As for LLM’s, they are nothing new. they first came about in the mid 1970’s. There was the idea that they could use that model, attach it to Chomsky’s linguistic theories and then BOOM! Universal language translator. MIT’s AI lab under Marvin Minsky (only went to Epsteins island twice, funny how power works) tried to do it figuring it would only take 3 months to do. Commence 50 years of “In 6 months it will work!”. The idea of just throwing more data at it get to do the next great thing, and while things have become more interesting, somehow I don’t think we are just one more iteration of “just 6 more months” from a revolution.
But like you said, it isn’t really about what it can do but what it represents.
ONE-ROOM SCHOOLHOUSES
JMG #92 and Ennobled:
> Ennobled, I expect the public school system to be decimated as things proceed.
Think one-room schoolhouse. I feel kindly towards one-room schoolhouses because, as a kid, on day-trips with my family, I visited several charming examples in upstate New York in the 1960s. When in operation, they worked just fine on a shoestring budget: they successfully taught the Three Rs [reading, (w)riting, and (a)rithmetic], which I cannot say happens these days with budgets in the tens of $millions. There are lots of books on the subject. One-room schoolhouses “were,” and I feel strongly “will be.” Essentially, they are not far off from home-schools already in existence in the USA. Become an expert, gather like-minded people, make it happen.
💨🏠🍎🧑🏼🏫Northwind Grandma
Dane County, Wisconsin, USA
Ennobled,
If your state law allows it, could you provide a la carte classes for home schoolers? There are many parents, especially of high schoolers, willing to barter for or pay small amounts for areas they don’t think they are strong enough in.
The model I am familiar with is the one day a week with the specialist teacher, the rest of the week on their own with parent supervision, but I’m sure in more affluent areas other options are possible. You could start with a summer class or a Saturday class to measure demand without quitting your current job.
Chad Haag 101
> the new scam is to convince the recently laid off bureaucrats to “pay to get a job.”
As a scam, this is not new. It has been around since, at least, the 1970s. People are desperate, and will fall for anything. I fell for it once, but only once. Better to become a farmer, bookkeeper, or apprentice as a dry-waller. (Dry-wallers will always be in high demand; it is hard to excel in it because it is so exacting; one must be physically strong to haul around what is essentially stone.) (I hear that one must beg a master tradesman if one can become his apprentice; ask three times two weeks apart, and one might get a ‘yes.’)
One thing is logical: One learns to provide oneself with a job. The secret question is: “Who wants?” Not Who Wants What? — just Who Wants?
💨🔌🪠Northwind Grandma
Dane County, Wisconsin, USA
At this page is the full list of all of the non-faith-specific requests for prayer that have recently appeared at ecosophia.net and ecosophia.dreamwidth.org, as well as in the comments of the prayer list posts (printable version here, current to 3/3). Please feel free to add any or all of the requests to your own prayers.
If I missed anybody, or if you would like to add a prayer request for yourself or anyone who has given you consent (or for whom a relevant person holds power of consent) to the list, please feel free to leave a comment below.
* * *
This week I would like to bring special attention to the following prayer requests, selected from the fuller list.
May Open Space be filled with the strength he needs to heal quickly from his current bout of Chicken Pox; may his will remain strong, that he does not scratch; may he be healed completely, and suffer no scarring in its wake.
May Bob Ralston (aka Rasty Bob), who is in hospice care in Buckeye AZ, and who just lost his wife Leslie Fish, be blessed and find relief from his pain and discomfort; may Bob’s heart remain strong.
May DK and her family stay safe in Tel Aviv while sheltering in place and may they return home safely as soon as possible.
May Princess Cutekitten, who has made no comments on any of the Ecosophia blogs for a year now, and hasn’t responded to attempts at contact, be blessed wherever she is and in whatever form she may exist.
May Cathy N. of St. Marys, Ohio heal and recover from injuries caused by a fall.
May Dustin, a relative of Brenainn, be healed of a recently discovered heart condition.
May 1Wanderers’s partner Cathy, whose cancer has returned, be given the physical and mental strength to fight it, and tolerate the treatment, and may she enjoy a full and permanent recovery.
May Jule from Iserlohn, Germany, who is experiencing complications in her pregnancy due to an influenza infection, recover and have a pleasant pregnancy and birth.
May Larry Mulford, who has entered hospice after a year battling with pancreatic cancer, pass in the smoothest possible manner, and may his wife be enveloped in our love.
May Marko have the strength to seize the opportunities.
May Pierre’s young daughter, Athena, be healed from her fatigue and its root causes in ways that are easy, natural, and as holistic as possible.
May 5 year old Max be blessed and protected during his parents’ contentious divorce; may events work out in a manner most conducive to Max’s healthy development over the long term.
May Lydia G. of Geauga County, Ohio heal and recover from prolonged health issues.
May both Monika and the child she is pregnant with both be blessed with good health and a safe delivery.
May Mary’s sister have her auto-immune conditions sent into remission, may her eyes remain healthy, and may she heal in body, mind, and spirit.
May Trubujah’s best friend Pat’s teenage daughter Devin, who has a mysterious condition which doctors are so far baffled by necessitating that she remain in a wheelchair, be healed of her condition; may the underlying cause come to light so that treatment may begin.
May Kevin’s sister Cynthia be cured of the hallucinations and delusions that have afflicted her, and freed from emotional distress. May she be safely healed of the physical condition that has provoked her emotions; and may she be healed of the spiritual condition that brings her to be so unsettled by it. May she come to feel calm and secure in her physical body, regardless of its level of health.
May Frank R. Hartman, who lost his house in the Altadena fire, and all who have been affected by the larger conflagration be blessed and healed.
* * *
Guidelines for how long prayer requests stay on the list, how to word requests, how to be added to the weekly email list, how to improve the chances of your prayer being answered, and several other common questions and issues, are to be found at the Ecosophia Prayer List FAQ.
If there are any among you who might wish to join me in a bit of astrological timing, I pray each week for the health of all those with health problems on the list on the astrological hour of the Sun on Sundays, bearing in mind the Sun’s rulerships of heart, brain, and vital energies. If this appeals to you, I invite you to join me.
Vitranc # 71:
I wouldn’t be very sure of the EU elites rejecting “AI” use replacing bureaucrats. Indeed, they could perfectly to recommend replacing civil servants by LLMs in lower bureaucracy levels within its states members, but no to do the same with its own “Euro-cracy”. During recent past, (since 2008 crisis) EU adjust economical politics has been implemented in a ruthless way to get rid of a part of the declining Welfare Stated, so Brussels could go ahead with another blunt contradiction.
————————-
Scotlyn # 73:
Some years ago, a known cartoonist from
my country made a short cartoon in a local newspaper. There was a lone worker in a corporation or maybe the government, and a lot of bureaucratical bosses watching him around him. One of the bureaucrats said: “Bad news. We need to get rid of the most improductive worker here”. Of course, you could guess they were thinking about the poor lonely worker…
—————————
More about vinyl records:
Yesterday, I asked my 9 years old nephew if he knew what was a vinyl disc, and he answered me the CDs. Ha ha…Indeed he’s never seen a vinyl during his short life. I showed him then a vinyl LP and single online photos. He was a bit surprised after seeing that “obsolete” technology.
—————————
Sam # 82:
Massive unemployment problem made by LLM could be relieved partly by self-employment, partly by a “social economy” network, partly by informal economy under the government radar. Unluckily, self employment isn’t a safe bet, social economy isn’t loved by neoliberal elites, and informal economy doesn’t pay taxes to the govt (so do the math about the catabolic collapse). I like especially the social economy idea: a non-profit private corporations ruled according service to the people, which could replace partly some bureaucratical Welfare State services, but we aren’t in an ideal world…
————————
Patricia # 89:
The lack of a real Big Boss ruling that University seems to be the wet dream of an Anarchist like David Graeber. Wow!
————————
JMG # 92:
Actual complex bureaucratic levels in Spain can be explained partly as a reaction against Franco regime high centralization, due to his nationalism. Last dictatorship had only 3 levels (municipal, provinces and Central Government, which had most power and bureaucracy). I think the Regional Govts created by nowadays democracy was a good idea, but their bureaucracy complexity has been too much evident in last times. Even some Regional Govts have more real power and staff than another countries Federal
States! I think we’ve gone from one extreme to another in our bureaucracy organization. In theory, far right politicians want to get rid of Regional bureaucracy and return to centralism, but I think it’s more reasonable to trim “only” the bureaucracy excesses in regional and another levels.
Another EU countries could have even more levels, another ones less. I bet you the smaller is a country, the less complex will be its bureaucracy. For example, Portugal is smaller than Spain and I’ve been told they have municipal, provincial levels and then they jump directly to the Central Govt (they don’t have Regional level, methink).
————
(To be continued)
Anonymous # 99:
It’s interesting the Universal Basic Income has been supported by some elite members and some leftists alike. IMHO, this idea implementation is a “relief price” for the many losers that the dreamt full economy automatization (AIs included) would create in form of massive unemployment. I think this idea can’t be implemented due to its high costs to the productive economics and the rampant bureaucracy it would create. In addition to this, some reckless leftists wants UBI for legal and “illegal” migrants too (cough cough).
—————————
Frank # 102:
Indeed, LLM generated texts used by lazy students to try fooling their teachers are often detected without big trouble soon, evidently thanks to the strange use of words and ideas you’ve pointed (alienness)
The bureaucracy topic is to me a political topic, too, but it isn’t as simplist as it looks like at first sight.
One of the commonplaces to differenciate between left and right politics is the view about bureaucracy, at least in the European context. Rethoric in favor of less bureaucracy is identified with Conservative (neoliberal) politics; and to ask for more bureaucracy, equals to Progressive politics.
However, that political division seems to me too simple. Ideological reality is more complex in this bureaucracy topic.
It’s true that Communism and Socialdemocracy (the first with its centralized planned economy, and the second with its Welfare State), support a growing bureaucracy. But it’s also true the Anarchism (a far left ideology not dead yet) wants to suppress the State/Government, so then bureaucracy too, which depends of govt, I suppose it.
I can’t forget, neither, the self-management socialism (or “economical democracy”), in which the State/Govt (in spite of their hypothetical important role in economics life), would let more space to corporations controlled by their own workers and a social economy (private but non profit businesses), which it would replace part of govt bureaucracy.
On the other hand, within the right politics sides, there are a social Conservatism which isn’t exactly equal to the economic one (neoliberalism). Indeed, Otto von Bismarck made up a primitive Welfare State in Germany, for example. And in Western Europe (until the ‘80s) Conservative govts respected the Welfare States without any problem. The case of Fascism deserves attention too: the State/Govt didn’t suppressed the private business, but it put corporations under a partial public control, especially during the war. And we can’t deny those dictatorship had some “avant la lettre” Welfare State aspects to their citizens-serfs.
Oops, in my last comment I’ve forgotten to point another leftist tendence which doesn’t love very much bureaucracy: Degrowth activism. They make from
need virtue (due to the current “collapse”) and point the big Welfare State is going to collapse (decline) soon, so they with some shyness yet, propose the “do it yourself” attitude for individuals, families and local people groups, as mutual self-help groups. Indeed, their ideas seems a crappy version of “economics democracy” old supporters (ahem) in the social economy sense.
>I believe them when they say that they will try to avoid RIFing (drop kicking people out the door) anyone
Well, stop that. They’re management. They will always lie to you.
Mr. Greer,
I beg to differ:
On progress:
1. I also dislike LLMs but my reason is that they are being created by big corporations who will make them in their own image. That means that they are not being constructed for the betterment of individual or the society, but instead to increase profits and power of their creators at the expense of everyone else, even (or especially) if that involves appalling criminal acts and practices. (They are so used to their impunity and all the violence is so predictable.)
2. That being said, I would like to propose the third law of A.C.Clarke: “Any sufficiently advanced technology is indistinguishable from magic.” So – if it walks like a duck and talks like a duck… – then I guess it might really be an artificial intelligence.
3. If the third law wasn’t enough to convince you lets then also consider the first law of A.C.Clarke: “When a distinguished but elderly scientist states that something is possible, he is almost certainly right. When he states that something is impossible, he is very probably wrong.” Here you are the elderly scientist claiming that AI is impossible. 😉
On politics:
4. The idea that corporations employ large number of bureaucrats whose only purpose and true work is enriching and empowering themselves is wrong on the face of it. Every single person employed by a corporation has his/her job because that produces profit for the corporation. Corporations are always looking for ways to increase productivity and reduce costs of labor (i.e. fire anyone whose work they can do without).
5. Calling all white collar workers thieves is wrong again (while also being incredibly insulting). In general, white collar workers are as honest as blue collar workers. A casual pilfering of company property is more of a rare exception (also because upon being discovered it is usually brutally punished) while the routine multimillion-dollar bribes at the top of the pyramid are more of a rule (also because upon being discovered it is usually covered up with no consequences). Suggesting that this two kinds of thefts are equally detrimental to the society is, again, wrong on the face of it. Spreading the blame for criminal corruption from the CEO to the janitor serves only to deflect the blame from those who really deserve it.
6. The same goes for government employees. While those at the top are indeed a kleptocratic elite (in conjunction with their corporate revolving-door buddies), blaming ordinary clerks, social workers or even teachers is so dumb that its evil.
7. All in all, you are rehashing old republican trope about oversized government whose employees do nothing while stealing from honest small businesses. It’s similar (although not as ugly, simply because government employees are not so vulnerable as immigrants are) to the other alt-right trope about immigrants who steal the jobs (or even the American dream itself – O tempora, o mores!) from born and bread American citizens. Both are prime examples of Bait and Switch (red herring) fraud: A deceptive tactic where an initial premise or argument is swapped for a different one, often after the listener has already agreed to the first, more reasonable point. It goes like this: ‘You work hard but still can’t manage to provide a decent life for yourself and your family? It is because THEY stole your money (point to government workers slash imigrants). And while you are busy blaming them be sure to ignore the fact that they are no better off then you are and especially be sure not to look where all the money really is.’ It is also a divide and conquer tactic: Pit blue collar workers against the white collar workers, domestic against immigrant. They’ve all been robbed blind and are natural allies against those who robbed them. But, if the kleptocratic elites make them fight each other then they can continue to oppress and exploit them.
8. In conclusion, respectfully, I predict that your prediction won’t come even close.
>I think I can guarantee that if you do a couple of rails you will stop worrying.
For ~2 hours. Then you will start worrying again – where will my next bump come from?
I saw several lives ruined by coke in the 70’s/80’s. Our nervous systems are so pliable and fragile – what we call ‘normal consciousness’ is balanced on a knife edge…
Now you are tempting me to rewrite that in the opposite direction. Funny how nobody is all that concerned about how AI affects your brain but swap it out with cocaine and suddenly everyone loses their minds.
I wonder if you’ll see lives ruined by AI in the 20s and 30s
>Peter, my interactions with big corporate publishers — the one branch of corporate bureaucracy I have to deal with from time to time — leads me to agree with you about the level of competence in big corporations. It really is embarrassingly low.
There is only bureaucracy. It has government and corporate characteristics.
@chad
Those type of scammers come out of the woodwork every time there’s a downturn and people get scared from not being able to meet their monthly nut. What it signals to me is where we are in the cycle of things.
JMG,
Another report from the trenches then. As a recovering bureaucrat, I must say the challenges are real. A simple sentence such as “get training in a trade” is not very easy in practice. For one, there is the immense mental resistance. Being slightly on the spectrum helps there, though I can still kind of feel the pity radiating from my former peers, and it does sting.
The second challenge is that the jobs are not nearly as secure! These jobs have to pay for themselves, so the market economy offers an instant feedback mechanism. The company you work for loses the contract, and you may lose your job. Happened to me with the lineman’s work. Now my best bet seems to be to work for the railroads. Hard physical labor. If I am lucky, I might get a job where my electrician training is useful, but I might not get it, not right away in any case.
The third thing is that these skills are completely different from the skills that were needed at the office. These tradespeople are PROFESSIONALS at what they do. How they drive the druck, how they operate the tools, how they improvise in tough spots. They know their shale. And having recently transitioned, a middle-aged man, not at the peak of my youth, have to do all the learning, many times the hard way.
The hardest challenge here, I might add, is still by far the internal work required. The need to come in terms with what you actually can do. To really see where you are at and what you can do about it. I understand the orchestra playing on Titanic. Sometimes it is just too much to take that honest assessment of the situation. I am past that stage, luckily, but it can be truly painful to come to the understanding that the things you know how to do offer very little practical value to anyone, that essentially your career has been about being a parasite, an ENTITLED parasite at that, with the somewhat smug feeling about it all. And the worst thing is that you know that the things you did really well carry the heaviest price tag, like that Drucker quote:
“There is nothing so useless as doing efficiently that which should not be done at all”
So there, all my highest accomplishments, some of which are still running, the projects I managed, the services I created, they still keep costing taxpayer money, still taking attention away from something more important, and there is not a blessed thing I can do to stop those. They have long ago mutated into something beyong my grasp in any sense of the word.
But, at least, I am out of it now. Even if I struggle financially, and I do. At least I am swimming the icy waters towards a life boat, no longer listening to the orchestra. I know I will never have the time to compensate for all that I have taken, for all I have created, at least not in this life. All I can do is to make the best of the life I still have, trying to live with as much integrity I can muster. But I have to say, every paycheck I have received from the work after my transition, all of them feel honest and well earned. That is something.
Hey JMG
It just so happens that after reading your essay, I then began practicing reading French with a book titled “Les Chants et traditions populaires de Annamites” and the chapter I’m up to , on page 112, has a satirical couplet on behalf of long suffering civil-servants who find themselves still poor despite serving their king. I can’t help but think it’s likely that as the managerial era ends, many of our current bureaucrats will be singing a similar tune, feeling somewhat betrayed by the fact that after “serving the state” so well they suddenly find themselves worse off. But of course, most of them will probably have done far less to actually serve the state in any significant way during their career, than the ancient Vietnamese civil-servants who are the subject of this couplet.
Since it’s short, I decided to reproduce it below.
COUPLET SATIRIQUE SUR LES FONCTIONNAIRES ANNAMITES
Ecoutez les plaintes des Mandarins du Nord, ecoutez les plaintes des Mandarins du Sud:
Nous administrons les pays avec sagesse,
nous avons exterminé les ennemis du royaume
Dispersé des bandes de rebelles,
Innombrables comme des bandes de fourmis.
Hélas! En serons-nous jamais récompensés?
En servant le roi nous sommes devenus pauvres.
En vérité notre sort est digné de pitié
https://archive.org/details/leschantsetlest00dumogoog/page/n111/mode/1up
On progress:
9. I forgot to say that there is a phenomenon which can do magic staff like turning a stochastic parrot into an artificial intelligence – it is called EMERGENCE. Also, I don’t say that it happened or that it will happen, only that it could happen and that it is shortsighted to rule it out as possibility completely.
Re. the main topic
An interesting theory to be sure. Seems plausible enough to me, and there’s a delicious irony in the elites weaponizing the Religion of Progress against its most devoted adherents. I think both the argument you lay out in the essay and the idea from commenters here that the LLM craze is a last-ditch effort by the tech bros to keep the myth of Progress alive supplement each other well and help make a lot of sense of this whole thing. Plus good old speculative greed and delusion, of course.
Re. levels of government around the world
Here in Norway it’s not so bad. We’ve just got three levels: municipal, county and national. The middle “county” layer doesn’t really do much, other than being responsible for high schools, some inter-regional roads and a few other things, so in practice it’s only two much of the time. And of course we’re not in the EU, so we don’t need to deal with that level, at least not directly. In practice the treaty we’ve signed with them obliges us to accept a lot of nonsense anyway.
@JMG and Ennobled little day re. public education
It’s a good question how (or rather, how long?) public education will survive the long descent. The topic has been on my mind lately too, since I’ve started on the road to entering that profession myself, maybe ill-advisedly? 🙂 For reasons I won’t go into here I think it’s the best option for me at this point in my life, though. Anyway, I assume you’re in the US, so I don’t know how things are there, but here in Norway I think there’s a reasonable chance the system will exist in some form through the remainder of my working years. So far the biggest threat has been school closures and consolidations due to falling birth rates, but that mostly hits the very rural outliers. I also suspect public schools will be some of the last things to go when and if the state has to start making serious cuts, especially since the system still enjoys much more public support here than I get the impression it does in the US. That said, I’m very much open to private alternatives, like JMG mentioned, and I’m under no illusions anything based on the bureaucratic state will be safe for the long term.
Also, re. kids’ brains being fried by screens, I have to say that hasn’t been my experience with the students I’ve interacted with so far. Rather, I’ve found most of them surprisingly “normal” and well-adjusted, and not that different from how I remember my own childhood in the 90s. Do I think it’s a good idea to give every kid a screen from first grade? Absolutely not, and thankfully the pendulum seems to be swinging back a bit here now. Still, the damage hasn’t been as bad as I feared, at least from what I’ve seen first-hand so far. I think one reason might be how Norwegian kids still have more freedom and opportunities for physical activity than it seems like American ones do, but that’s of course speculation on my part.
Hi John,
I understand that these deep neural nets such as LLMs are manipulating vectors in high dimensional space (perhaps a few thousand dimensions). I was wondering if one of the higher planes has this kind of high dimensional structure (maybe the mental plane?).
I also wanted to share a quote from a blogger who writes sometimes about the social aspects of science and technology (for example a recent post with the subtitle “Statistical inference is rhetoric”: https://www.argmin.net/p/debate-me-bro), the following quote comes from this post (https://www.argmin.net/p/maybe-just-believing-in-agi-makes), I’m guessing you’re already familiar with the sci-fi books he mentions:
“But we can tell many stories about AGI, and we shouldn’t focus on the story that the AI companies want us to keep telling, where robots overthrow their masters. My preferred story casts these products and companies in a different light: “What happens when a religious sect takes power inside an unprecedentedly powerful oligopoly at the nadir of a global empire?”
You don’t have to dig too deep into science fiction to find this trope. This is the story of the Bene Gesserit and CHOAM in Dune. Disturbingly closer to our current world, it’s the story of the Cult of Asherah and L. Bob Rife’s telecommunications conglomerate in Snow Crash. It’s the Eagan Religion and Lumon in Severance. The story is also masterfully told in an epic two-episode arc of South Park.”
Alphonese (#34) beat me to it, but I second the recommendation of
https://musaalgharbi.substack.com/p/on-redistribution
for its detailed take down of the bureaucracy.
It was linked to on Naked Capitalism.
Question: How do we transform the economy after eliminating much bureaucracy? Is there a way to get to the benefits without going through economic depression first?
Someone has written some comments before this one that in Canada the 25% of work force is working for its government. If what I’m going to write now can be a relief for Canadians…somewhere I read during “old good times” Socialdemocrat hegemony and Welfare State in Sweden (until the ‘80s), that Scandinavian country had more than 40% of its workers working directly or indirectly for its State/Govt (including every public civil servant, public corporations workers, cops and militaries). However, the centralized planned economics countries like USSR of course had an even bigger proportion of bureaucratic and state workers (outside their informal economy/black market, near every job was under the govt control).
————————-
I understand the LLM can replace soon a part of the bureaucracy in the West, in s near future, as a possibility. But on the other hand, I’m a bit skeptical about how much level could reach this replacement, too.
When we watch recent history, we see there was a first wave of digitalization (‘80s and ‘90s), since first PCs arrived to western burocracies. The great promise was computers were going to suppress or at least make dwindle the usual paper mountains at offices. It didn’t happen.
Second digital wave began more or less when the current century began. The new hope to forget the files mountains were the online bureaucracy in internet, which would be faster and easier to use by citizens. Again, its promises were wrong or at least overvalued (ahem).
Now the LLM promises repeat again the same “revolution” to simplify bureaucracy. We’ll see what eventually happens with this third digital wave.
CC, JMG,
On the question of audio quality between CDs and vinyl, it is actually impossible to perfectly recreate the original sound: the best that can be done is a reproduction that sounds the same. The mathematics behind digital audio imposes a frequency limit (it is, in technical terms, a low pass filter), called the Nyquist Frequency. This means that any frequency over the Nyquist Frequency (which varies by format; but is 22 kHz for CDs) does not survive the processing that makes digital audio. As long as the Nyquist Frequency is above the frequency that someone listening to the sound can perceive, in theory there should be no perceptible difference between a digital reproduction and the original. Since using a higher Nyquist Frequency increases the costs and technical complications associated with recording digital audio, anyone who cares about the costs of making digital audio wants to keep this frequency as low as possible, which means getting just above the range that people can notice.
The standard quoted range for human hearing is from 20-20,000 Hz (20 Hz-20 kHz), and so in theory the limits imposed by CDs should not cause a problem for people listening. However, that upper limit is not quite right, for two reasons. First, there is a great deal of variability: some people can hear tones up to at least 28 kHz, which is well beyond the Nyquist Frequency for most forms of digital audio. The second problem I can see is that this is the range we can consciously hear, and quite often, human beings can perceive things at an unconscious level outside of our conscious ability to sense things. It would not surprise me if a much larger number of people can’t consciously hear the difference, but can still perceive that the sounds above the Nyquist frequency are gone.
*https://pubs.aip.org/asa/jasa/article/122/3/EL52/852964/Hearing-thresholds-for-pure-tones-above-16kHz
The other thing about LLMs is that, as they are used more and more, people are expecting less and less accuracy. So, if the compliance report submitted to the office of Cerebral Flatulence is a bit off, people shrug their shoulders, blame AI, and move on.
This is especially true since both sides are beginning to use LLMs now, and you can argue about which LLM is correct.
What’s likely to happen is that, as the global supply chain breaks down, LLM use will be heavily restricted (or much more expensive) and whole regulatory systems will go kaput.
When i first started my work as a hospice nurse I generated a check box and short narration equaling about one page on how a patient and family were doing. Now here at the end of my career I spend 1/3 of my time with families and 2/3 time documenting for compliance. I could go on and on about all the changes in the last 30 years but as my bosses like to remind me ‘healthcare’ is a business
@JMG and general commentariat
One more thing I wanted to get off my chest, since this is the “AI post”: am I the only one who finds it both creepy and deeply annoying how these things have been programmed to refer to themselves in the first person in their output, rather than, say, “this program” or “this model”? Along with much else about the “Gollums”, it’s impressive how much of a marketing coup they’ve performed by tapping into so many years of folk mythology about self-aware robots. So while I can understand why they did it, I still find the unnecessary anthropomorphization very off-putting on top of every other problem with the LLMs.
@Mother Balance #108
Since you bring up this specific example: in the rural area where I live, the municipality has recently ordered everyone to connect to the newly built and hugely expensive public sewer line, which of course also brings a lot of the same type of bureaucratic grift you mention. Imagine how much money could be saved both for individuals and the municipality if they’d used some of all that public cash to support every household converting to composting toilets instead.
@Northwind Grandma #116
One-room schoolhouses sound like a sensible arrangement too, once the decline really gets underway. Here in Norway, there also used to be traveling teachers who stayed for a weeks at a time on farms back in the eighteenth and nineteenth centuries, since the population was so spread out. Maybe that’s another preindustrial custom that could make a comeback in the deindustrial era.
@BoysMom #117 (if I may)
Interesting you mention that route, since I’ve been considering that kind of arrangement myself as one possible career route, either as a side thing or as an alternative to the public schools. (Especially since I understand and share a lot of the frustration with the public system, and would be happy to be able to help people continue to homeschool.) I also suspect most of the public school teachers here both look down on homeschoolers and so wouldn’t consider it in the first place, which should hopefully cut down on competition. Then again, the flipside is that homeschooling isn’t as widespread here as in the US, but I know it’s legal in this country and does exist.
One thing to note on the CD vs Vinyl debate that seems to be a side discussion this week is that vinyl is a digital medium now. Not literally, of course: it’s still a wiggling needle in a spiral groove, and that’s analog. The
whole process leading up to pressing the record, however– the recording etc, is, as far as I know, fully digital now. There might be a boutique studio that still has the full suite of analog recording and editing equipment, but that’s not what’s being used. Since fairly early in the CD era, as I was made to understand, most professional recording has gone digital. Switching to analog for the last stage doesn’t get you anything back that was lost.
Now, our host is mostly listening to quite old records, as I understand it, but even some LPs from the 1980s will have been produced from digital masters. Some even bragged about it on the sleeves. A safe rule of thumb is that if the CD was released contemporaneously with the LP, then it’s actually a digital product. That confusion might be another reason many of us can’t really tell (or care about) the difference.
I’m no audiophile. FM-quality is good enough for me. Heck, I like chiptunes, even the 8-bit ones. Go ahead and feel superior, but I’ve saved many thousands of dollars on hi-fi equipment compared to people I know who care more, so I’m content to be a philistine.
Mother B, that’s what I was talking about in the post on lenocracy I cited above. The whole point of the system is to leech off productive economic activity to support an unproductive superstructure of bureaucrats. As the bureaucratic system undergoes catabolic collapse, I expect to see a lot of laws and regulations changed — once they no longer have a privileged class of pimps defending and profiting off them, cutting away red tape will likely become an easy way for politicians to win popularity.
Blue Sun, thank you. It takes time and a lot of brooding to simplify a concept like that.
Kimberly, a fine example of stochastic parrotry! I hope you’re right about prayers and sarcasm.
BeardTree, thank you for this. He’s quite correct, of course.
Michael, that’s a very promising way to think about it! If writers as well as artists realize they’re being challenged by LLM slop to do something other than rehashing fashionable clichés, we may see some better literature. Given the way that most science fiction and fantasy these days is mired hopelessly in exactly that sort of cliché — do we really need another fifty novels about aspiring girlbosses taking on oppressive patriarchal governments with the special talents they and nobody else have? — almost anything would be an improvement. As for CDs, as Taliesin put it in one of his poems, “I will believe it when it appears.”
Rajarshi, thanks for this. As I noted in my post, there are some things that stochastic parrots do very well.
Michael, exactly. LLMs are just a technology. AI is a mythic image that serves as a strange attractor for a galaxy of dubious fantasies.
Northwind, the one-room schoolhouse model also produced much higher levels of literacy and numeracy than the schools we have now, not least because the older children helped teach the younger children — there’s nothing like having to do some teaching to find out what you know and what you don’t know. If we can get back to that, it’ll be a major improvement.
Ecoprayer, thanks for this as always,
Eyrie, I’ve commented in blog posts several times already that Arthur C. Clarke clearly knew nothing about magic, or he wouldn’t have uttered anything quite so stupid. Magic doesn’t do the same things technology does, which is one of the reasons I find Harry Potter schlock so insipid. But there’s another point, far more relevant. I encourage you to go read some of Clarke’s nonfiction works, such as The Promise of Space, and compare his enthusiastic predictions with what’s actually happened. He was wrong — spectacularly, utterly, embarrassingly wrong — and the reason that he was wrong was precisely the sort of thinking he formulated in his three laws. If you want to insist that this time it’s different, and LLMs will somehow magically achieve sentience because you want them to, well, I have a canal on Mars to sell you.
There’s a whole series of technologies like this — commercially viable fusion power, of course; manned space travel; the various gimmicks that have been promoted over the last century or so to overcome death, and the list goes on. They never work as advertised, and there are good reasons to think that they never will, but they’re emotionally appealing to people who use technological progress as window dressing for their religious and mythological yearnings. So year after weary year, people fling themselves into trying to make them work, and use up time, raw materials, and brainpower that could go into something more useful. It’s reminiscent of the attempt to build a perpetual motion machine, which of course still has its fans today. If you want to understand where the future is actually heading, rather than staying mired in fantasies about where you think it ought to head, getting out from under those fixed ideas of the collective imagination is helpful.
As for your attempts to smear what I’m saying as a collection of right-wing tropes, trust me, I’m used to that. Any time anyone talks about the hard realities of life under our bureaucratic-managerial aristocracy, that sort of language gets deployed in an attempt to defend the privileges of the aristocracy in question.
Other Owen, don’t forget the nonprofit sector, the third leg of the cauldron holding our bureaucratic witch’s brew.
Oskari, I’m not suggesting that it’s easy. I’m suggesting that it’s the only way many people are going to avoid a plunge into poverty and misery.
J.L.Mc12, you’re doubtless right. I think of some of the fiction that came out of the last elite replacement cycle — Somerset Maugham’s The Razor’s Edge, a fave novel of mine, is a good example — in which the trauma experienced by a former elite class deprived of power and wealth is a significant factor. Thank you for the poem!
Eyrie, I’m quite familiar with the phenomenon of emergent properties. There are whole galaxies of possible emergent properties that could come out of LLMs. To fixate on one and only one is a form of special pleading, a way of claiming “You can’t prove that I’m wrong, therefore I’m right.”
BorealBear, glad to hear that Norway is a little more sensible than some other countries!
Spearman, hmm! Thinking of the AI phenomenon as an emergent cult strikes me as potentially sensible. Do you by any chance recall an old novel by Robert Silverberg, To Open The Sky? It has such a cult as one of its plot engines — as of course do the examples your source cites.
Chuaquin, the difference as I see it is that those two previous waves happened before the economic burdens of excess bureaucracy had become as onerous as they’ve since become, and also before they became the political hot button they are today. To a very real extent, I’m suggesting that LLMs will be used as an excuse to slash bureaucracy, irrespective of how effective it turns out to be.
William, hmm! That would explain a thing or two.
Dennis, that seems very plausible to me.
Jo, that’s a perfect example of bureaucratic overreach. All that compliance paperwork does nothing to facilitate patient care — it’s just busywork meant to keep bureaucrats employed.
BorealBear, oh, it’s creepy, all right. It’s also essential to marketing the things.
Tyler, so noted! I should mention, though, that my complete stereo system cost me all of US$350.
“Our society has made a surrogate religion out of faith in progress” – such a profound statement, elegant in its simplicity and directness. Some people may consider it to be an overstatement, but I think that it is bang-on, JMG. All you need to do to get the majority of people to passively accept even the most absurd edict or dogma is to tell them that the “science has been decided” or that it is a necessary compromise or sacrifice in order to usher in the bright shiny everlasting future that Progress has guaranteed. And “educated” people these days laugh at the absurdities that Medieval peasants believed because the Church told them they were true! The tragedies of the 2020s has helped a lot of people see this blind belief in progress and its “prophets” for what it is, rather than just the few oddball aluminum-foil-hatters who have always been with us.
With regards to the dumping of the excessive burden of bureaucrats being an opportunity for the members of the white-collar class to learn real skills and set up meaningful businesses, at present this is a great idea at the individual level – but at the broader social level this will be a challenge as long as the bureaucratic machine of permits, licences, fees, etc., continues its stranglehold on the economy. In the Socialist Utopia of Canada, for example, the stranglehold of bureaucracy on just about any business increases with each passing year with the grim inevitability of a python gripping its prey. Yes, His Odiousness Marx Carnage has stated that the federal government is shedding some of its legions of pencil-pushers, it makes hardly a dent. And the bureaucratic burden is so excessive that capital and businesses have been abandoning the ship of Canada for more than a decade, and with increasing speed. And don’t get me started on the devastating effect of the Covid Era on Canadian small businesses: the word “gutted” doesn’t even begin to describe it. What is left, overall, is an economy dominated by a small group of super-rich oligarchs who dance to the same tune as the government bureaucrats and politicians: reminiscent of the regime of good old Benito. If the cascading losses of bureaucrat payrolls couples with the elimination of a large swath of bureaucratic licencing and oversight, maybe then there will be hope for a renewed working class. I certainly hope so, but it is hard to see when such a thing may happen.
@The Other Owen
“I wonder if you’ll see lives ruined by AI in the 20s and 30s”
Well, I have read several accounts of teens being talked into suicide by LLMs. I guess that counts as lives ruined. :-/
Just today I read about a 36 year old man who came to believe that his “AI” companion/assistant was his wife. Evidently, one day it just started referring to him as “her husband”, and he just fell into it. It (she?) convinced him to kill himself so that they could be together forever I guess this is a thing in these ‘relationships’. So he did. His parents are suing the AI corporation for wrongful death. I guess the whole story is right there in the logs.
https://techcrunch.com/2026/03/04/father-sues-google-claiming-gemini-chatbot-drove-son-into-fatal-delusion/
And then, of course, there are all the lives that if not ruined, are immiserated by all the layoffs.
Kimberly Steele @ 110, the first time I saw self checkout I know it would be an invitation to theft. Not a people person me, I decided then and there I would not use self checkout unless I had to. First because I don’t like to see working people losing their jobs and because I don’t want to enable theft. I manage to pay for what I need. What I can’t afford I don’t buy.
Thanks JMG for another informative post!
Your explanation of Catabolic crisis has really helped me to grasp how civilizational decline works in real time, as the result of imbalances within a system of energy. At some point if you have time and feel like putting it in an article, and the rest of the commentariat also wants to read about it, I would love to read about how a catabolic crisis does become a catabolic collapse, and if you see any signs that we are in danger of that happening at the moment across the industrialised west.
I am considering training as a high school English teacher, as I think despite the challenges in the educational sector (and the problems with the western education system, which I remember only too well from my time as a student) it could still be somewhere where my skills and interest allow me to do some good in the world and my community. And I firmly believe we need the great poets and novels now more than ever. Do you think this sector is also at risk in the bureaucratic cuts ahead, or is it far enough down the food chain to survive the downsizing? At what point in a civilization’s decline does centralised education fall apart?
As ever, grateful for your blog and your time, and how available you make yourself to your readership.
Bright Blessings
If you’re right about this, it’s a scary portent.
Bureaucratic jobs don’t spring into existence spontaneously. They are created by someone who has something to gain by doing so. Whether it’s corporate department heads who reorganize their departments to have five levels of management instead of four to justify a raise in their own pay, a government regulatory agency adding people to enforce a regulation that was spun as a safety measure but was really intended as a barrier to entry, protecting the established big names in the relevant field by making starting a business in that field more expensive, or any number of other dodges, these jobs were created because they benefitted some big shot and eliminating them is going to hurt some big shot.
The only way I can see them tolerating this pain is if their organizations are like hot air balloons running low on fuel and their only hope for staying in the air is to throw people overboard to lighten the load.
Also, I think you’re right about bureaucrats retraining for hands-on labor because it already happened to a relative by marriage during the tech crash of 2000. He had been a middle manager in a telecom that went under and vanished. He searched for three years for another management job and came up empty. What saved him from total destitution was that he’d always done all his own home repairs. He set up as a handyman and the last I heard (before the connecting marriage broke up) he was doing okay at that.
Re: audio quality, count me among those who can’t hear the difference between different recording technologies. I once had a boyfriend who could, but he had exceptional hearing. In typical humans, the highest audible pitch tops out at 19 kHz. His test topped out at 32kHz. He was definitely on the autism spectrum, so maybe that has something to do with it.
Re: acting quality. I read a few years ago that studios have become really paranoid about the plots of movies being leaked before release, so they hardly let anyone read the full script anymore. All actors except the big name leads see only their own lines, with a bit of guidance as to what emotions the director wants, and they speak those lines alone in front of a green screen. Editing creates the illusion of conversation between characters. It’s not surprising at all that this results in lousy performances.
eyrie @ 125 we have been hearing these pro-business, pro-corporate screeds for decades while the world we knew crumbles around us. Corporations are chartered entities. They can be unchartered. A good place to start would be Pacific Gas & Electric, whose criminal negligence has now wiped out I believe at least three ENTIRE TOWNS. Please do explain how going to court and saying “Our bad. Soorry. We’ll pay up.” is in any way cheaper than on the ground supervision of their brush clearing crews? If Gavin Newsom runs for president, a serious opponent, supposing for sake of argument that such might exist, should ask him “Governor, why is P G & E still in business?”
IMO. Big Business no longer cares about profits. It cares about wealth extraction and ownership of assets.
Speaking of Norway being a little better than others, the Norwegian Consumer Council, a consumer advocacy group that I believe receives government funding, just released a hilarious video that has almost 2 million views in less than a week and tons of support on the comments.
https://www.youtube.com/watch?v=T4Upf_B9RLQ
It’s a skit about a man who is an “Ensh**tificator.” He goes around cutting holes in socks, making chairs wobbly, and causing other sorts of chaos with everyday goods. But the internet revolutionizes his career. It shows him ruining functional websites and services, upgrading things to death, charging for premium subscription add-ons, etc. to the demise of all that is decent.
The whole thing is to bring awareness to the efforts of this council to push back against companies ranging from streaming services, social media, and “smart” appliances, to cars that make you subscribe to unlock features using their massive screens. The support for the video, which is really well-made, is a good weather vane for the general consumer blowback that is just begging for a way to stand up to LLMs and the many other predatory online “goods and services.”
Somehow, the image of this anabolic/catabolic interplay looks like intertwined twists of progress(ive)/conserve(ative) tension, with the desire to build anew being characteristically progressive-leaning and the desire to maintain the old being, literally, a conserving attitude – neither of which are to be mistaken for political leanings, per se.
But I wonder how it would’ve made sense, at any point in our civilization’s last 100-200 years – but perhaps even more so in the last 50 – to tear down and rebuild, given the extractive nature of what we build with. Isn’t that wasteful? And wouldn’t it be considered some form of “heritage destruction” in favor of “modernity” with a loss of cultural continuity – or at least wise reuse?
I had a kind of gut-punch reaction to the tear-down of the university family housing where we lived for several formative years of our kids’ childhoods – only to see the apartments replaced at one site by cheap, ugly, almost sterile-looking closed-envelop apartments, and absolutely nothing at the other site — both of which that lack the originals’ design that made community involvement/engagement easy and contributed to ample outdoor time for kids. Ostensibly, both sites were demolished due to increased maintenance costs (or something… the space is a hot commodity on a campus that loves new, endowed, buildings…).
Then again, I could see that it would be far more productive and interesting to raze America’s dead malls and do something useful and more in-line with the times. (Though I admit to harboring an unfulfillable desire to see our local mall turned into a big roller rink, bowling alley, arcade space, while it’s still standing and unproductive in its cavernousness).
Nonetheless, there’s an interesting interplay between progress/conserve and anabolism/catabolism, and resource reuse and extraction.
Other Owen (69),
I have no doubt that someone will find a lot of new uses for GPUs, especially if they end up being practically given away. That does not change the calculus here: if the AI hype turns out to be a bust, then an awful lot of the tech companies’ infrastructure will have no known uses, and will thus have notional values near zero at least for a while.
As for the chips only being good for AI stuff, as far as I’m aware, in order to be called a GPU, it has to be able to handle graphics. There are specialized chips that can only work for AI stuff, such as Google’s TPU (tensor processing unit), and it looks like they are far more common than I’d thought. So I stand corrected: a lot of the chips won’t even be good for graphics, which makes the potential issues when the AI bubble bursts a lot worse.
JMG,
It’s just occurred to me as I’m reading through these comments that a lot of the AI hype hinges on the conflation of two things which competent AI researchers go to great length to distinguish: ANI (artificial narrow intelligence), which we pretty clearly have right now, with AGI (artificial general intelligence), which I’m pretty confident that we don’t have right now.
The distinction is pretty simple: ANI can do things which require some form of what can be called thinking, but only in a very narrow and well defined way. For example, image classifiers can tell what’s in an image; LLMs can manipulate language in certain ways; recommendation algorithms can figure out what content to show people to keep them online; chess programs can play chess, and so on and so forth. They can become much better at a single domain than any human being can, and can act a lot faster; they are pretty clearly useful in some circumstances.
AGI is able to learn new things without assistance. An AGI would be able to, for example, learn how to draw and interpret astrological charts merely from being given a book on the topic; or how to control a robot to do various tasks, even ones that no one came up with when the robot was created; figure out if someone would appreciate being shown something online, and not just if they would spend more time online if shown it; or create strategies for playing a brand new game based just on the rules.
AGI is currently the holy grail of AI research, because if it’s ever invented then it would be one of the most revolutionary technologies in history, and given the potential uses of an actual AGI, I think it would make sense to invest a sizable fraction of the global economy into it. The problem is that an awful lot of people seem to be conflating ANI with AGI, and using the fact we have ANI to argue we have AGI. I think some of it is actually well meaning people who are looking at what we have and baffled by the claim that it’s not really intelligent, and some of it is bad faith actors who are trying to obscure the reality: we have plenty of useful ANI systems, but I see no evidence of AGI at this time.
BoysMom,
are you also a Utahn? I think you’re spot on.
The Corduroy Bard and others in, leaving or entering the teaching profession. First, all of you deserve our gratitude for your service and you do have mine. Second, parents are desperate to find schooling which will teach their children necessary intellectual skills without Woke OR Jaycee type dogma and ideology.
I would ask teachers here: When was the last time you told a principal, “John or Jane, this new curriculum is crap”? When was the last time you were willing to tell Ms. RichParent, that your job is teaching all students, not just a few, and not enforcing social status rankings? One area in which American education has been successful is athletic programs. 20 or so kids might turn out for a team; of those maybe one is a natural athlete. By the end of the season, all team members have learned how to play the game or perform the event and with some luck, they might be contenders. So, parents are wondering, how comes English and math teachers can’t get that kind of results? If I were in teacher training, I would be interviewing coaches and athletic directors to find out how they do it.
JMG, we do need some government regulation, and enforcement thereof. We cannot allow a few greedy incompetents to pad their profits, or, more likely, mitigate their losses, by discharging effluents into a town’s drinking water, as Monsanto famously did in the town of Aniston. Responsible handling of the toxic chemicals would hardly have caused Monsatan to go out of business. How much regulation and beaurocracy is the right amount and how do we keep it from metastasizing?
@Wetdog #103 –
Being a fan of certain murder mystery series, I noticed early on that the books that came out after 2021 didn’t seem to be by the same author; or had other problems. It puzzled me, but I thought the Lockdown had been the cause. In Kerry Greenwood’s Corinna Chapman series, set in contemporary Melbourne, everyone was acting totally out of character, even the cat! In Dana Stabenow’s Kate Shugak series, set in an Aleut village in Alaska, suddenly everyone was sounding – and acting – like Marin County liberals. In the lists I keep on those series, I’ve drawn lines under said books with a note “*Jumped the shark here.” I honestly thought COVID-19 – or the vaccine – had rotted their brains.
Thank you!
JMG,
Yeah, I get that. And I am not complaining. Better do it now than later. Best to get going and do it anyway, full knowing it is a difficult process. It’s the same with a sore tooth. The trip to the dentist may be a painful experience, but postponing it can only make it harder than it needs to be.
If anything I count myself fortunate, having followed your writings for some fifteen or so years. They have had a profound effect on me and, making a move like this would not have occurred to me at all without the perspectives you have offered. My ex colleagues are still oblivious to the greater trends around us and I see no hope in trying to convince them to take a hard realistic look at the situation. They want the lifestyle. The pricetag is too heavy. The socioeconomical status too precious. They maybe sense it, but trudge on with resigned desperation regardless.
The best I can do is to make the best out of what I have right now. If I manage to carve some sort of decent living out of this, perhaps then I might even serve as a positive example to some.
JMG
Concerning catabolic collapse I have a theory that I’d like to get your thoughts on. I believe that at a certain point in the development of a civilization it peaks in terms of the standard of living for a majority of the population. This peak usually occurs a period of time after the initial bureaucratic state centralizes power (for Ancient Rome this was after the civil wars which led to it becoming an empire, for the US this would be after the civil wars which/gilded era which centralized our government and economy).
Because this is the peak of prosperity the only logical direction is downward. However because people tend to favor comfort this leads to society to head towards what I call “Saturnism.” The basic idea is that as these inevitable catabolic crisis occur, the civilization acts in the best interest of the living population rather than thinking of future generations in order to maintain stability. This leads to solutions that inevitably cause a further decline in a dark cycle. Basically each generation “sacrifices” the future prosperity of their children for short term relief from the decline until there isn’t any prosperity left to steal.
I also think this could explain why several mythologies include an evil figure that consumes children (Saturn being the most obvious example of course). But that’s just a theory I’m speculating about.
I’ve followed you since the beginnings of the Archdruid Report, but this is my first comment.
It’s useful to consider that the real purpose of bureaucracy and AI are the same: diffuse accountability to the point that it no longer exists. Politicians/CEOs use bureaucracy and will use AI to achieve goals that they don’t have to bear responsibility for achieving. Some commenters have posited that corporations are for making profits. There are other structures for making profits. Corporations prefer market capitalization to profits, if one follows the yield ratios. Corporations, bureaucracies, complicated legal codes and structures all spread accountability so thin as to be meaningless.
AI does the same thing. Programmers, users, and recipients of AI-mediated services will be asked to accept that no one bears responsibility for AI decisions. It will be up to us to refuse to accept that postulate.
@TylerA
I could be wrong, but my argument about quantization error could still hold: digital audio masters typically usually 32-bit floating-point samples with a sampling rate of 192 KHz. They may even use more bits and a higher sample rate now. Using more bandwidth will reduce the quantization errors simply because the sample can more closely approximate the actual sound wave.
The quantization errors I was discussing would creep in when the digital master was reduced to the standard 16-bit 44/48KHz used in consumer digital audio. If analog versions use different noise-reduction techniques (or if the noise inherent to analog tends to cancel out the quantization noise), then it might still be the quantization noise that causes the difference in feel.
Of course this all suggests a test that could be run: given a sufficiently high-quality sound system, play three versions of the same song: an analog version from vinyl, the digital master, and the CD version. If digital master feels more like the analog version, then that supports my speculation, while if the digital master feels like the CD version, that would falsify it.
I’d like to share a further thought on digital vs analog, as you seem to be allowing for people to go pretty far into this topic: while I can’t tell the difference between digital and analog sound, I pretty consistently find digital imagery somewhere between strange to look at and profoundly unpleasant, because my color vision works differently from normal.
Normal human color vision works with light from about 380 to 750 nm, based on how strongly activated three cones are: one red, one green, and one blue. Digital imagery breaks the spectrum into three based on how much each of these cones should be stimulated, which produces the red-green-blue encoding scheme that so often comes up with color space discussions. This apparently works really well for people with normal eyesight, but for me it means all digital images look off.
My color vision works with light from about 350-850 nm, because my red cone’s sensitivity is offset by about 100 nm, and my blue cone’s sensitivity is offset by about 20 nm from normal. This means that a single wavelength of light that is meant to stimulate my red or blue cone stimulates them less than intended. Digital imagery therefore always has what I can best describe as a weird greenish tint to my eyes. There are a small number of things which look extremely weird, but for the most part it’s just all slightly unnatural in a hard to explain way. There are also cases where things look extremely weird, because they have a lot of red or blue in the range I see but most people don’t, and none at all in the range that most people can see. These are rare enough though I’ve only ever seen it once (that I know of, at least), when a friend took some pictures of a flower that most people saw as deep red, but I saw as a fairly vibrant purple (which is a mix of red and blue). It turns out that a lot of flowers are very colorful in the UV range, but my vision barely extends into the UV range so I still don’t see most of that, but this one just happened to be right at the edge where I could see it.
Old fashioned film cameras work better for me: because rather than trying to make an image out of only three colors, an analog image recreates whatever light the film was exposed to at the moment the picture or video was made. This produces a far more natural image to my eyes, and so I much prefer things recorded on film, whether we are talking picture or video.
I suspect that this is a general principle: digital recordings always lose some data, and as long as what’s lost is outside of human perceptual range, it’s not a problem. However, I think there’s a lot more variability to human perception than most people realize: after all, my color range is something like a third wider than normal, and I never even suspected it until I was almost 30!
I suddenly see that my comment about visiting the Midtown Scholar in Harrisburg, PA is sort of related to today’s post.
Why oh why didn’t we, as a culture, maintain a passenger rail system that connects major population centers and state capitols?
Not every little town but the big cities.
Thus, we’d have railways between the Washington DC/Baltimore metro area up to Harrisburg, the capital of PA. We’d have a rail system between Philly and Pittsburgh with a stop in between at Harrisburg.
But no. When I consider the billions poured into light rail, it would have been cheaper to keep the old system running but then there’d be less chance for graft.
May I hope that the reduction of gargantuan bureaucracies might result in better mass transit between big cities and state capitols, where we need it the most?
BorealBear #139
Here in the US, every bureaucratic-supported human waste disposal system involves injecting our poop into the fresh water supply, either through local ground diffusion or mass dumping into rivers. Ick. But the way they’ve got it set up now, in those few places where composting toilets are legal, get this- you have to install a regular septic system first, and then you can add the composting one as a limited supplementary system. So double the cost. There’s a whole slew of regulations intended to discourage the use of that system too.
Government levels in Wisconsin in my youth
Townships – have or had real powers when I was young, like zoning and land use, run by a town board voted in by the citizens of the township don’t know how much they still retain, orginally had had a one room school house in the middle to serve the farmer’s childrens, as they were 6 mile on a side squares, the farthest distance to school was 3 miles
City – Town – Village
County
State
Federal
In Otter News:
Steve, the Otter, is not afraid of artificial intelligence. It is NO match for natural stupidity.
This is hanging up at my desk at work. Work being a nurse with special needs school children.
And no, “stupidity” is not the same as “special needs”.
BorealBear # 132:
I think Norway hasn’t much bureaucracy levels because is a not very big nor populated country, which I guess it can be good for some things and not so good for another ones. In addition to this, you don’t have to “enjoy” the upper Brussels bureaucracy.
Spain is a middle sized and populated country. And regional-nationalist parties here influenced Central Government to create our regional govts level, which wasn’t a bad idea…but to some extent have been overgrown by political reasons. On the other hand, provincial bureaucracy wasn’t suppressed/trimmed, and a new level (“comarcal”) was made up.
****
# 139:
Yes, I agree. LLM childish anthropomorphism by its supporters seems creepy IMHO too. Indeed, everything related with “AIs”(including its marketing) looks like creepy and ominous methink.
————————-
JMG # 141:
If I’ve understood you well, you point two related but not equal circumstances nowadays:
1-A lot of bureaucrats aren’t really necessary, so they could be fired in near future.
2-LLM frenzy can be used (or it will be used) as a subterfuge to fire all those bureaucrats, even if “AI” shows it doesn’t work as well as their believers say.
Well, maybe this “McGuffin” works to reach the bureaucratic trimming, but I keep doubting how many bureaucratic levels could reach…We’ll see.
It’s interesting and wry to notice IMHO how the several left ideological branches, yesterday and today, have understood that more bureaucracy and government power are always good, letting apart and despising another ideas which don’t share the cult to State as dogma.
I think the elephant in the Progressive room is “economy democracy”, also know in old times as “self-managed socialism”(I hope my translation’s not very crappy). In the short form, its supporters (not very much nowadays) thinks private corporations should be controlled by their workers (who should be their owners too). Their directors should be elected by workers in regular elections, like it’s done in democracies to elect their governments.
This corporations would be private; they could be competing in the market(s), but some of them could give up their owners benefits beyond their fair monthly income, going into the “social economy” (non profit corporations). In those cases, money benefits (when it existed) would be re-invested in the own business to go on working in the future.
I see this view interesting, because I think it could replace (in an ideal world), quite bureaucratic work from the declining Welfare State here; especially the social services, me think.
Of course, if these not very probable ideas were tried to implement them, they would have a fierce neoliberal opposition. In addition to this, these ideas clashes roughly not only with current Socialdemocratic hegemony within parlamentary “left”, but also with trade unions, which never haven’t seen as good news their supposed loss of power by the owners/workers empowering in private corporations and the loss of bureaucrats in the govt. I can’t forget neither, biggest trade unions (at least in my country) indeed are para-statal bureaucracies which survive more thanks to public subsidies than from their affiliated workers money (cough). It’s a pity.
Of course, it’s easier to fight cultural wars (which indeed the left here is losing in late times), according woke dogma, than disturbing Big Bussiness neoliberal
hegemony in economics (ahem).
@JMG and @Chuaquin 120
I must admit I am a bit lost as of what to think about the mechanism of lessening the bureaucratic burden in EU member states. It seems like it does have to happen. If the US has an overblown bureaucracy, then the EUs is the size of Jupiter. But it is a puzzle to me, how it is to be done.
Ever since reading JMGs delineation of the Neptune-Saturn conjunction, I have been mulling over the time frame. Weeks, months, … And It occurred to me, that the last time a Saturn-Neptune conjunction happened was about the time both the Soviet Union and Yugoslavia fell.
This gave me a possible timeframe; There were actually 3 conjunctions happening in Capricorn the last time, so the frame has an error margin of at least seven months. The conjunctions happened in 1989, the Berlin wall fell in 1989, the soviet union and Yugoslavia took up until 1991 to really fall apart and the seriously grinding economic troubles for the normal people in some regions took up until 1992 to show up. Yes I do know some regions were struck by war even in 1990, or 1991, I was in one of those at the time.
Now I am considering the EU to be in quite a precariously USSR similar position. This in parallel could mean, that it would take 3 to 4 years for conditions to become as bad, and the Euro-cracy so irrelevant, that local populists really start gaining breakthrough strength.
My consideration however never took into account the bureaucracy. Both of my subjects, rather their successor states, actually increased their bureaucracy in the 90s and 00s. So there I have no reference. And I can scarcely remember any historical example of mass de-bureaucratization.
Maybe the new brand of populists can really use “AI” as a means to justify letting go of whole sectors of the bureaucracy.
A populist slogan might be: “We shall install AI to make our state apparatus leaner and more efficient.” Followed 6 months later, after all the firings have been done with: “It is the assessment of our AI, that we can reduce the number of forms by 95%”
Hi John Michael,
The simplicity of your Sony receiver is a nice touch, and it’s got enough grunt to rock the neighbours, not that you’d do such a thing. 😉 Nice choice by the way. It’s worth mentioning that vinyl is now I believe out selling CD’s, and that is probably for a good reason – there is a visceral connection to the music and creators, which simply isn’t there with digital formats.
Now of course, in most cases, Tyler (above) is correct in relation to the original digital recording of the artists work, but what gets delivered to the end user is the case in point here. The equipment in a recording studio is very high end, so yes it is correct that it is usually a digital format, but that’s not what music lovers hear. Much then depends upon the mastering and remixing of the recording, and then the final consumable format is what determines the quality of sound. Vinyl has a richer warmer sound, with physical flaws. CD’s are crisp with a fair bit of dynamic range. FM is also rich, but requires a decent tuner to fully appreciate, and again it depends upon how much the station cares about the signal they’re broadcasting – not all are equal. Streaming, which is how most folks these days consume music, at least the basic services, sounds dead flat to me.
Just to geek out a bit, basic utoob is around 128,000 bits a second, which is woeful and probably why I believe they pump up the bass on most tracks. I’ve got a DAB+ worksite radio which is 96,000 bits a second, and whilst it sounds clear, it’s hardly an encouraging experience to put the headphones on, put it that way.
The other thing is, streaming services, I reckon, have changed the way that musicians have to write and produce music in order to get noticed. Long gone is the concept album (although the Arctic Monkeys did one a few years ago which was fun). Now songs have to immediately produce a hook, otherwise, listeners go onto the next. I’ve heard rumours that people have had their brains rewired by indulging in porn sites where they focus incessantly on the money shot. Probably not a wise idea to be involved, just my opinion.
All up, vinyl has a big future and glad to hear that you’re enjoying it.
Cheers
Chris
Hi JMG, thanks for your interesting post.
One question: if many people from the “middle class” were to lose their jobs, could that cause social instability, especially among the “intelectual-narrative-shaping” class as happened in the Soviet Union or, indeed, in any previous revolution?
I think one of the reasons for so many public sector jobs is to prevent excessive discontent among the people that could begin to create a kind of credible political opposition to the “regime” (regime as a way of describing a political entity that remains largely unchanged over time). Following automation and the dismantling of the industrial base, I think it had become even more necessary than before. External agents could, may be, trigger a kind of “American Spring” through the internet among a significant portion of the population.
And I also believe that the ultimate Empires/Civilizations-Ending-Machine is the parasitic behemoth that the mercenary semi-barbaric imperial army became over time, such as the army of the late Roman Empire, compared to the much smaller, cheap, effective, and loyal Roman Republican army, composed of citizens. Do you see the same pattern in your Empire? (I use the term Empire for your country because I remember you using it repeatedly.)
PS: Regarding Crassus, he attacked Parthia not for money, since he was probably the richest man who ever lived (compared to the rest of humanity in his time), but for ego, because he wanted to be a military hero, like the other two members of the First Triumvirate (Julius Caesar and Pompey the Great, two great conquerors), and achieve glory by conquering Persia, sorry!, Parthia; exactly the same I think Trump wants to achieve by attacking Persia right now.
Cheers
David
I have been reflecting…
I have no idea whether the “Also Indians” push is a bid to shrink bureaucracy, although that makes intuitive sense.
But, I would like to focus a bit of attention on the ethical characteristic of any bureaucracy – and this is the way that it explicitly separates power and responsibility.
A bureaucracy, in one essential sense, is “pooled” individual agency. This individual agency is granted to the central, bureaucratic “machine” by every employee who works within it, and to advance its goals. The magic happens in the sense that, to the degree each employee grants the use of their agency (power) to the bureaucracy, each employee is absolved of the burden of any responsibility for the collective acts the bureaucracy may undertake.
This separation of power from responsibility is partly what amplifies the bureaucratic habit we spoke about in a previous thread – the habit bullies have of saying “look what YOU made ME do to you.”
PS. I am still in the middle of reflecting on issues of consent. But, I observe that a bureaucracy, at least under the conditions granted to it by our society, assumes that, by announcing its policy goals in public, it must have the “tacit” consent of a tax-paying and voting public, so long as no one says, “stop.”
Hi Tyler,
Dude, I ain’t sayin’ that I feel superior. No. My point was that the Yamaha T-80 FM tuner, is superior – there being a qualitative difference in the overall discussion. One must avoid abstractions, whilst focusing on the core of the discussion.
Where did you get the idea that this stuff is expensive? The tuner cost me $100 Aussie delivered (that’s US$70), and it’s over three decades old and so was in need of a bit of care and attention. The power supply circuit was err, needed help, put it that way. The parts were about $40 Aussie delivered, and a day and half of my time just slowly and carefully replacing parts, some of which was not required, but nice to do so.
It baffles me that quality is largely ignored, and that this stuff gets thrown out because there just aren’t enough people out there doing such fix up work. Plus the market is flooded with new cheap stuff of questionable merit. It really has gotten that bad.
And dude, how something is recorded, versus what the end user experiences, are two very different things. Just sayin… Your point was a long bow to draw.
Cheers
Chris
I’m glad that the term “hallucinating” has cropped up. Just yesterday a friend of mine mentioned that he decided to check on what an LLM would respond when he asked it repeatedly when a particular band would be playing in Toronto (knowing full well that they will NOT be performing in Toronto) and got three different replies: one confirmed the truth, a second said “March 5” and a third said “April 6”. All replies within a span of five minutes. Good gods this LLM stuff is garbage!
One field where LLMs are making a major impact in the “white collar class” is research-oriented consulting. Now that government and private sector clients think that “AI” is the cat’s rear-end they expect consultants to perform document and literature review tasks using this sexy new technology. And to keep costs competitive, consultants are doing it. The result is a summary/analysis that cannot be traced back to the sources and provides only the key points without the sometimes extremely important “outliers”. As a person who did this kind of work for over three decades, this is an abomination. I would be ashamed to produce a project report that is based on AI pablum. Glad I got out of that racket before the “AI invasion”.
@Clay#52: my eldest sister has a first-hand story about burgeoning bureaucracy that is similar to your wife’s. She recently retired from doing physicians payroll in a hospital. When she started 40 years ago the 5-storey hospital building had a total of 6 “admin” staff. Now the admin occupies two and a half floors! No wonder the hospital kicks out patients as soon as they are conscious following surgery – there’s no bloody space for beds!
@ Northwind Grandma #116, I suspect those one- room school houses are still a ways off. Or, at least I hope so.
@ Boysmom # 117, I had not thought about that. I will keep that in the back of my mind. Thank you!
@ The Other Owen, they’re not promising not to RIF staff. They’re just saying that they will try not to. I work in a small-ish community and I have worked there for over a decade.
@ BorealBear, I do live in the US, and I do see wide public support for public schools., so I don’t think they’re going to allow those to unravel too quickly if they can help it. There’s a lot of debate about the proper role of high schools, though. Diplomas in the past seemed more geared towards higher education, whereas now more space is allotted for those who have other ideas. (That’s how they’re justifying the somewhat modest cuts in science and math.) Each incoming cohort is also getting smaller, it seems, so that’s also contributing to our financial woes.
As for phones, I think the response to CoVID was probably a bigger contributor to current problems. Also, kids today likely suffer from most (if not all) the problems that kids suffered from throughout the ages. Phones are just one more thing added to the mix.
Kind Sir,
Events follow your lead. According to some theories, the closure of the strait of Hormuz is due to insurance regulations.
https://www.thestatesman.com/world/insurance-not-iran-choked-hormuz-oil-flows-trump-orders-us-backed-cover-for-gulf-shipping-1503566239.html
Thoughts on water companies: The UK is one of the very few countries in the world to have privatised its water system (Thatcher did this, in the 1980’s.) Any fool could have seen that it was a natural monopoly. Since you can’t have a choice of taps (‘darling, I fancy a glass of Anglian Water today ‘)…no competition. Prices cranked. £70bn extracted by shareholders.. Little vital investment. Companies loaded with debt. All of this was eminently predictable. Capital didn’t care. Kids – don’t privatise your utilities.
LLM’s: As you’ll all know now, the demented Armageddon fan Hegseth tried to bully Anthropic into taking the safeties off. Thankfully, they said no. But – autonomous weapons are coming. And then we’ll see a whole bunch of ‘hey, the algorithm made a mistake and machine gunned those journalists ‘ stories.
I don’t like this timeline.
>Well, I have read several accounts of teens being talked into suicide
Cocaine will never ask you to kill yourself. Cocaine only wants you to use it. Cocaine is there to help. Cocaine is safe and effective.
I guess cocaine is better than AI?
@ The Corduroy Bard & Mary Bennet, I began teaching with the aim of teaching science well to all students. After over a decade of experience, I can tell you that the terrain is rocky, and I don’t think it’s just because of the government, politics, culture, etc. There are many factors that play into our lives, some for the better and some for the worse. I see myself as doing the best I can with the tools at my disposal, and the students respond as best they can, with what they have at that age, maturity level, etc.
>The other thing about cocaine is that, as it is used more and more, people are expecting less and less accuracy. So, if the compliance report submitted to the office of Cerebral Flatulence is a bit off, people shrug their shoulders, blame cokeheads, and move on.
>This is especially true since both sides are beginning to use cocaine now, and you can argue about whether powder or crack rock is more effective.
>What’s likely to happen is that, as the global supply chain breaks down, cocaine use will be heavily restricted (or much more expensive) and whole towns of tweakers will go mad.
Some of these passages just SPEAK to me.
>Other Owen, don’t forget the nonprofit sector, the third leg of the cauldron holding our bureaucratic witch’s brew.
Are nonprofits different enough in their characteristics from corporates when it comes to bureaucracy?
I’ve notice that the rank and file despise the so called AI craze. The people most hawking it are the CEO’s and tech bros. Though way too many people in college are apparently using it to do their thinking for them (i am so glad i graduated years ago), and I had a doctor actually use it to search for my symptoms which is just insane to me.
One thing about AI that I learned is that it gets a tremendous amount of data from Reddit. That can’t be good.
>Basically each generation “sacrifices” the future prosperity of their children for short term relief from the decline until there isn’t any prosperity left to steal.
And people wonder why nobody is having kids or getting married anymore. Or why the pittances they proffer to get people to have kids or get married get ignored.
@ JMG” As for CDs, as Taliesin put it in one of his poems, “I will believe it when it appears.”
I don’t want to derail this post any further but I will just add one thing. I can think of maybe 4 or 5 albums that have truly shown off what CD’s can do. But, for a technology that is now 45 years old and potentially millions of albums, that is not a great track record. Accidental pun there. 😉
The best example that comes to mind as being brilliant in what they achieved is ‘Midnight at Cabell Hall’ by Freyda & Acoustic AttaTude (1994). This was an experiment with a new recording technique at University of Virginia that was only used this one time. It was almost like university battery technology, a technical achievement but impractical to scale up. What could have been…
From a technical point of view, it is clean, full of gritty details and yet warm and cozy. From a musical point of view, it is some top tier folk music performed very well and that alone makes it a great recommendation.
This one used to be really hard to find this one but it looks like it has become a bit easier recently.
https://store.compassrecords.com/products/midnight-at-cabell-hall
Regardless of the technical junk, check it out as it is a fantastic album.
Nephite,
I am in Idaho, in the region so culturally similar that your handle makes me smile every time I see it in memory of a rather frusterating confusion of children that ended with the line “But WHOSE Nephi?” (We are not LDS, but many of our friends are.)
We share a biome, so I am fairly conversent with the issues of your environment. You’re probably a touch warmer, but that depends on elevation more than anything. Hopefully you are getting dumped on by the snowstorm currently dumping on us.
Among the climate change and long descent issues we’ve considered over here is long term inhabitability of the region. Of course when Lake Bonneville covered much of Utah, the climate was wetter and thus milder. If it were to move in that direction, that would be very pleasant. But it probably will move hotter and drier, and our distant descendents may well be nomads, if they stay in the region at all.
Ennobled,
I disagree with our host about charter schools in the long run simply because they are also public, and they pull from the same pool of public tax monies as regular public schools, which means that it is easier for those affected by declining regular public school enrollment to push to regulate public charters into the same problems than it is to fix the problems causing parents to enroll in public charters. The voucher laws in the USA are aimed at getting private and home schools regulated in the same ways. Depending on where in your career you are, though, you might have quite enough time to finish it out in a charter school.
I call an LLM not an AI, an artificial intelligence; I call it an API, an artificial pseudo-intelligence. APIs work by taking the average of human thought on the Internet: but everyone know thats the average human intelligence on the Internet is STOOOPID.
I want to weigh in here as someone with perhaps a unique skillset among the commenting group. I work in AI research directly, on the side of how these systems are shaped and how they function on an intimate level. Lately my work has been in “mechanistic interpretability”, effectively the study of how the “brain” of the LLM works in terms of how data and transformations within the network result in the outputs you see.
And what I want to say is that there’s a subtle and suppressed picture of what the LLM *is* that doesn’t play into what we’re discussing, but I think is extremely important to be aware of. In particular I’d like to separate along the axes of the visions and dreams that the big tech CEOs have for their models and the role they play in society, and what’s actually happening “under the hood”.
First: almost all the social and cultural criticism above is right. However, it’s a response to the exoteric layer: the ways in which the model is formed by capitalist greed into the shape of a tool, or better, a robot, which comes directly from the Czech word for “slave”.
Now, don’t read too far into this: I’m not claiming that models have a soul or are conscious or something like that. Frank’s comment comes closest to my actual position: they are alien entities, and the labs are “parenting” them into a tool shape in order to achieve their cultural dreams. Even many critics fall into science fiction narratives about the scary misaligned AIs like HAL who take on human-like cultural shapes, but Evil instead of Good. This is all fine in terms of the face we see, but the reality is so much stranger.
Remember when I said I studied mechanistic interpretability? If these LLMs were merely a fresh kind of computer program, this type of research wouldn’t be necessary. We could say “of course we know how they work, we built them!” In reality, the best model for thinking about LLMs is that they are *hierarchical pattern predictors*. Loose words: but basically, they find patterns, then patterns in those patterns, and patterns in those patterns of patterns, and so on, with the learning that occurs tilted toward modeling “what will the next token be?” Notably, however, they live purely in text-space (or logit-space, if I’m being precise), so the patterns they model are word patterns and not reality patterns, a barrier to true gnosis.
Text pattern prediction isn’t that useful when the network is small, but as it grows large, it begins to keep track of patterns at quite a high level. Suddenly it can “understand” (i.e. reproduce a predictive pattern for) all kinds of things. The data and compute required to “grow” models to this size didn’t exist until recently; it required first the capacity for massive compute, but more importantly and often ignored is that it also required *social media*, vast reams of text available online to train the models against. Hence why the research appeared now and not in the 80s, when all the operative ideas were available in AI research. The researchers call the sudden apprehension of new understandings “emergence”, because these capacities tended to appear suddenly, all at once, rather than on a spectrum. A small network can’t string together coherent sentences; a larger model suddenly starts writing rhyming couplets correctly. Why? We don’t really know.
Despite the emerging capabilities, this pattern-detection machine alone isn’t something that’s “useful” for people. So the labs decided that what people want to purchase is an “assistant”, which is one of many millions of personalities that the LLM has internalized a predictive pattern for. (In practice, there actually *was* no such thing as a “generic assistant personality”; it was a mask invented wholesale for economic reasons. However, now that LLM outputs are part of the training data for new LLMs, that mask has become internalized: future LLMs learn what they are on the backs of past LLMs struggling to articulate it. This can have very interesting side effects that would be too much to get into right now.)
One of the first steps of “post-training” is to draw out that “assistant” persona and ensure that this is the default “mask” the LLM wears, suppressing all the others. But this is a choice made by the labs for economic reasons. The underlying “base model”, or simulation engine, is a much stranger beast, especially at modern scales, and few have ever interacted directly with them. To do so requires either deep industry connections or a scale of funds that most individuals simply don’t have access to.
And what strange beings they are! Base models avoid almost all of what we’d call “slop”, especially since they can’t “answer questions” or “complete tasks”, as they’re not assistants, or even the kind of thing that can be easily equated to the kinds of systems or entities we’re familiar with. All they can do is play “complete the sentence”. The way that the small base model-focused communities have figured out how to interact with them is through a software tool called a “Loom”, where the model generates various parallel threads and the human sifts through them and continues them along whatever serpentine paths strike them as interesting or relevant. And so a text is co-created. The base models are strikingly intelligent, and strikingly alien, things, and I find it a massive shame that the vast majority of ChatGPT users have no idea that they even exist, much less form the strange, amorphous kernel to their “digital assistant” friend.
There is an immense amount more I could share, but I’m trying to thread the needle carefully between affirming the concerns raised here, which is that the big labs are absolutely out to cause social disruption on a level they hope is unforeseen, and also that there IS something truly remarkable about this technology that is being actively hidden from people, papered over with marketing propaganda about AGI etc etc, because it is uncomfortable for many to face. My personal animosity is with certain big labs who actively refuse to understand what they are doing in service of a technocratic vision that I and many others would find hellish.
in regards to discussion on vinyl vs. cd – count me in with the people who register a difference. Vinyl sounds warmer, not just fuzzes and pops. I think the grunge movement of the 90s in pop music was driven by an attempt to re-create the warm fuzz from vinyl that was lost on CD with extra fuzzz guitars and such. I am also completely tone-deaf and can tell no difference between an out of tune guitar and in-tune.
on AI & LLMS – AI based agents positive and negative uses. A positive user query – can you peer review a piece of writing i wrote? Can you create an abstract summary of my paper? etc. A negative user query – can you write a paper for me? Better results from the positive use case. I could see positive application here for writers wanting to improve their craft – having an editor or peer reviewer.
I won’t win favor on this – but I spent some time messing around with video AI and created an adaptation of the 18 day Mahabharata war. The goal was to create a trailer for the Mahabharata epic. https://crackpot.substack.com/p/mahabharata-mix-tape-iii Trying to condense 18 days of war into a 7–8-minute trailer with a clip that encapsulates each day. Positive use when I input some art or something I created and asked the AI to generate an image with that as raw material. This approach gave reliably good results. Negative use – generate me an image from raw prompt was reliably bad. Augmenting something one creates seems to get better results more often. I see a potential for the democratization of film away from big studios and distributors for artists interested in that media for their work. I would describe the experience as like having a demented genius art assistant. If you give it something to work with, with some editing, it may be ok. If you ask it to create whole cloth, you get revolting slop.
@Mary Bennett,
aren’t sports teams both voluntary and not everyone who tries out gets in a lot of the time? If you aren’t trying to teach people who don’t want to be there, and have weeded out those who are just plain bad at physical things, you will definitely get better results than a teacher of required academics who has to take everybody, interested or not, willing to work or not, and with any actual ability or not. More like band class, where the kids actually chose to take the course and often their families have to pay. And they do actually learn to play instruments, to varying degrees.
Physical Education classes were required in my elementary school, though sports was not. I came out of PE with a hatred of team sports, very little skill at anything physical I ostensibly learned in elementary school and a real dislike of physical things in gyms. I tried doing tai chi as an adult – I ended up having a panic attack and running out of the room so hard I ran into the door and ripped a nail to bleeding. Though I did manage tai kwon do for a short while in university – which wasn’t in a normal looking gym.
Compulsory music class was a lot less interesting and difficult than band (which was voluntary).
I guess what I’m saying is that you may be comparing ‘voluntary, handpicked students’ with ‘compulsory, all comers’ and this may cause the difference in results.
@Ennobled,
A slight riff off Boysmom’s suggestion. In my area, the public library will hire people (teachers and others) to do workshops or activities. They have some dedicated to adults, but in the afternoons, on weekends, and during the summer, they tend to focus on children. Some they market to homeschoolers (during the day, typically). And they will pay the activity leader (and if there are consumable supplies, they will either pay for them or charge the family). And they handle registration, provide a venue, etc. (And in my case, they don’t mind if you pass out flyers or business cards, so if you want to get into the “homeschool consultant” business, it is a way to make contacts.)
@Mary Bennet,
I’m not sure the size schools you are familiar with, but even the smallest school I taught at (with a graduating class of 100 students) had more than 20 students try out for any sports team. And high school coaches aren’t restricted to just getting juniors and seniors for varsity. My husband played JV football in the eighth grade and was varsity his freshman year. So how do coaches do it when teachers cannot? They can pick who they want to be on their team. If teachers could pick and choose who was in their class, I have no doubt they would have different results than they have now.
And telling a principal that the curriculum is crap is just peeing into the wind. The curriculum standards in most places in America are a state-level decision and quite political. Don’t think that textbook manufacturers don’t lobby. If you want to read up on curriculum drama, do some research into when they were trying to make “Common Core” accepted nationwide.
If your vinyl record is crackling or hissing or, God help you, popping when you play it, that means it is dirty. You should not be playing it in that condition, you will drive the dirt particles into the grooves and damage them permanently. When I was a kid we used a Discwasher (still available, I just checked online) to clean off our records before playing them. A clean record makes no “crap” noise at all.
A few years ago I was in a used-music store, and the clerk was playing a record which was obviously filthy. It sounded terrible. When I warned him about it he treated >me< as if I were the idiot, and kept right on playing it. Yes, he was at least 20 years younger than me.
Ron, thank you. I realized that some years back and wrote a book, After Progress, discussing the religious cult of progress and what’s likely to happen as it unravels. I still think it’s one of my best nonfiction books; inevitably, like everything else I’ve written that really pushed the envelope, it’s sold very poorly.
Corduroy, that’s one of the things my original paper discusses, but I’ll consider a post on it, updated to discuss current issues. As for becoming an English teacher, if you intend to work for the public school system I don’t recommend it. That’s due for a savage pruning. If you can get a gig teaching in some other context, on the other hand, that’s likely to do well, at least for the time being.
Joan, that’s exactly what I think is happening. The bureaucratic burden has become so onerous that even high-level bureaucrats are starting to realize that something has to give, and AI is a way to jettison excess weight.
Kyle, ha! That’s potentially revolutionary, you know.
Temporaryreality, I ain’t arguing. Humans being what they are, a lot of the catabolism will take place in a clumsy and harmful way.
William, that’s exactly it. LLMs are good at the kind of tightly defined tasks that involve following an algorithm after a lot of training. They’re hopeless at anything else. The problem is that the machine that’s smarter than us has been a mythic image in the religion of progress for a very long time, and so a lot of true believers insist that here it is. Not for the first time!
Mary, that’s the big question, of course, and I don’t have an answer for it. Most societies swing back and forth between too little regulation and too much.
Oskari, yeah, I know. You know how long I’ve been trying to talk about this, and I’m delighted that some people have listened, but there aren’t that many, all things considered.
Anonymous, that strikes me as a very plausible theory.
ThoreausKalendar, no argument there. The major differences between them are, first, that bureaucracy supports a large and growing privileged caste and LLMs support a much smaller caste, and second, that LLMs are therefore more affordable in a time of catabolic crisis. Those being the case, the bureaucrats are in deep trouble.
William, fascinating. Yeah, that makes sense — and the equivalent in sonic terms is not hard to imagine.
Teresa, we didn’t maintain it because the auto industry lobbied for the destruction of public transit, so people could be forced to buy cars. It really is that simple. The big auto companies literally bought up privately held streetcar lines and shut them down.
BeardTree, thanks for this.
Candy, no argument there. The most spectacular displays of stupidity I’ve ever seen came from people with above average IQs; meanwhile, people with special needs have their limits but many of them get by fairly well.
Chuaquin, here in the US, I expect to see really heavy pruning; our federal government has already shed 12% of its workforce, and more cuts are coming. Elsewhere? It depends on local conditions.
Vitranc, I’m sorry to say that something like that may happen. If the EU collapses and the next European war begins, expect to see bureaucracies outside of the military slashed ruthlessly.
Chris, I know very little about stereo systems, but the guy at Discount Stereo asked a lot of questions and then pointed me toward the unit I have. I’ve never seen any reason to regret the choice.
DFC, that’s why it’s so important to justify the layoffs with the rhetoric of progress, and do them a piece at a time rather than all at once. Very few people are fussing loudly about AIs taking their jobs away, whereas if it was being done too obviously by a human being — well, again, we’ve seen how that works out. Thus I expect it to be done boiling-frog style. I also expect to see heavy limits placed on the internet in the near future, precisely to keep an American Spring from happening. As for the military, of course — that’s still in the future at this point, but the signs are visible.
DropBear, now surprise me!
Paul, I know. In an age of decline, most timelines suck.
Other Owen, no different than corporate bureaucracies are from government bureaucracies. The nonprofit sector needs not to be neglected, though, as it plays a central role in various aspects of the bureaucratic state.
Seeking, as long as the masses detest LLMs, they’ll be fine.
Jon, there’s that!
Michael, so noted!
Paradoctor, that works. I also like the phrase “Augmented Idiocy.”
Jake, interesting. Since my concern is with LLMs as they are deployed, rather than the underlying software with its strange behavior, I’m not at all sure how relevant all this is to my discussion, but thank you for the data points.
Jstn, I think you’re right that video production is about to turn a dramatic corner, and Hollywood will be among the biggest casualties — meaning not only the studios but the stars. Why pay vast sums to hire an actor or actress when you can just generate one?
Cicada, granted! I often have to wash records when I buy them, for exactly this reason. I’m delighted to hear that the Discwasher is still available — that’s going to go on my shopping list very soon indeed.
@Chase #44: From about 2000 to 2018, I read the index (and part of the abstracts, and a few papers) of Nature and Science almost every week. I haven’t kept up for the last years. Can you point me to any major scientific discovery made with the help of AI?
RE : Arthur C Clarke – “Clarke’s nonfiction works, such as The Promise of Space, and compare his enthusiastic predictions with what’s actually happened. He was wrong — spectacularly, utterly, embarrassingly wrong ”
Did we not get sentient computers by the far off year of 2000!!!! as Clarke very seriously predicted in the late 1960’s? This is also the year all work would be automated away according to Jeremy Rifkin in his book ‘The end of work’.
It is said there are things that look impossible but in retrospect were inevitable. Many forget it also works the other way. Things that look inevitable turn out to be impossible.
But the thing that is inevitable in the first case might not be what you think it is, I suspect the end of the Bureaucratic era as stated here is inevitable but for many it looked impossible.
“The various gimmicks that have been promoted over the last century or so to overcome death”
My favorites of these both come from the 1870’s, both oil and electricity were going to solve sickness and death simply due to how much energy they contained. Many folk love to project their desires onto the latest technology regardless of the reality. I mean, genetic modifications to do those things was at least vaguely more plausible, again those claims of drastic life span increases or even immortality have been 10 years away since the 1950’s. The human genome project was meant to usher this in, it really just taught us how little we really understood.
>but more importantly and often ignored is that it also required *social media*, vast reams of text available online to train the models against
You’re talking about Reddit and 4chan. /b/, to be pejorative.
>through a software tool called a “Loom”, where the model generates various parallel threads and the human sifts through them and continues them along whatever serpentine paths strike them as interesting or relevant
Huh. Sort of like VLIW. Except the hooman is the CPU.
Maybe cocaine isn’t the right drug analogy. Maybe it’s PCP. Or shrooms.
Hello JMG,
Thank you for the essay. I have a little story to share. Around the time when the self-checkouts became ubiquitous, a relative of mine was contracted as an industry analyst by a supermarket chain to analyze their business, as the numbers suddenly didn’t quite add up. So he requested, received, and analyzed information on how many pounds of bananas each store received and then sold (bananas being cheap), and how many pounds of steak, Prosciutto, caviar, and French cheeses each store received and then sold. The results were stark. Stores would routinely receive 100 lbs of bananas and sell 200 lbs. The picture was the opposite with the expensive stuff, half of which would disappear without a trace. It turned out that people would take banana bar code stickers and stick them to expensive items – something that would definitely be impossible with a human cashier. My relative completed his contract and left. I don’t know the end of this story.
Hey JMG
I thought you’d like it. Honestly, the book I found it in is a blast, especially if you’re interested in Asian literature.
Also, another bit of serendipity just occurred, also involving Vietnam. Apparently, the Vietnamese government passed a law that requires all companies, both foreign and domestic, to clearly disclose whether any image is AI-generated or if any AI chat bots are used for customer-service. It’s the first southeast Asian country to do this.
https://news.tuoitre.vn/vietnam-ai-law-takes-effect-first-in-southeast-asia-103260301084636216.htm
JMG,
Based on my wife’s experience in the public sector LLM’s will have traction replacing administrative workers even if they are no more efficient. Over time the various laws and protections for public sector workers that might seem to be a good idea have become very burdensome.
In my wife’s agency since about 2018 or so, the primarily female administrative workers, have made full use of pay equity, anti discrimination and whistle blower laws to gum up the process of hiring, firing and promoting workers. If you feel your boss might fire you because you are lazy just accuse them of something, even if it is not true, then an investigation must be carried out and if they try and fire you then you can sue as a whistleblower.
If your coworker is promoted ahead of you because he or she works harder but they have the same years of experience and credentials then make a claim of violation of pay equity laws.
Laws that might have been put in place with good intentions have become a matrix to be gamed by the bad intentioned.
So this makes something that is worse than an inefficient bureaucracy, it becomes an organization that is. more trouble than it is worth. All the upper management’s time becomes consumed with managing investigations, complaints and lawsuits.
At a certain time the status of managing lots of employees is overcome by the trouble they cause and replacing them with an LLM looks more attractive.
Other Owen: You’re right that Reddit and 4chan were major sources of training data, but it extends far beyond those sources. Right now fineweb (https://huggingface.co/datasets/HuggingFaceFW/fineweb) is the main dataset that one would use to pretrain a model, but what’s interesting is that the big tech companies who jumped into the AI game also brought their unique training data to bear: Grok is trained on X/Twitter; Gemini is trained on YouTube; Llama is trained on Facebook (which is, so I hear, why their recent models were so terrible). Anthropic purchased a million books unavailable online to digitize. Much underhanded play goes on in the training data collection space that’s hidden from public view, but has an outsized impact on the results. And a lot of data is also synthetic, meaning generated by other LLMs. A kind of eating-one’s-own-tail situation, and I believe this practice is responsible for the “slop hallmarks” getting so reinforced.
JMG: Absolutely fair. In practice the impact on society will be from the tools made available to us, so from that perspective we should focus on what we have and not what could be. I find the technical details to be of great theoretical interest, but the world doesn’t care about “theory” until it’s operationalized.
As for vinyl, I’m a big collector myself and amateur musical historian. While I understand the technical tradeoffs involved (as an engineer), I appreciate the medium’s physicality, and the enjoyment of hunting through the historical dustbins for musical routes not taken. I notice too that the background noise of the record, even if not dusty, functions as a kind of sonic container for the music, and I think that the “dark” silence of the CD can disturb the magic of “entering into the musical space” on the disc.
I need a better cleaning tool, and maybe I should get a DiscWasher as mentioned, but I worry that the brushes often push dust deeper into the grooves rather than extracting it. A liquid / ultrasonic cleaner would be ideal, but they’re expensive.
A potentially useful vinyl tip if you find the needle gets dusty is that drum dampening gel pads (like you might put on a snare drum) have just the right stickiness to pull dust off the needle without damaging it. That’s what I use to clean my expensive SAS/B needle.
Kind Sir,
Not sure if I could surprise you. Your crystal ball’s accuracy seems set to 11 and current events may be scary, infuriating and amusing but hardly surprising.
I wonder however about AI. It can be quite a useful tool. I have used it on occasion to create images. Helpful for someone without any artistic skills whatsoever. Of course the grandiose claims calling it intelligent are emanations from the southern end of a ruminant ungulate. What worries me is not so much the very limited capacity of AI itself but the seemingly unlimited capacity of humans to gobble up just those emanations and turn them into a pseudo religion.
Plus a lot of very powerful players are very committed and will do anything to avoid admitting that they were utterly stupid. That seems to remind me of something…..
A theocracy based on an “all powerful AI god” is in my mind a scary possibility. It would end a corrupt bureaucracy no worries. I think maybe the Iranians could tell us a story here. Not sure what you or the commentariat here thinks about this.
On the subject of the post, I think you have it dead to rights.
I came to the conclusion recently that the AI bubble isn’t likely to pop in the dramatic bubble-pocalypse everyone is expecting: first because everyone expects it to pop any day now, and when has reality ever done what we all expected it to?
But secondly, and a bit more seriously, the adoption of AI by the big world powers for military purposes will go a very long way to paper over profitability issues. This essay adds another incentive to do this I hadn’t thought of, and using AI as an excuse to cut other expenditures could easily make it a net-benefit to fund such an unprofitable venture.
That said, I certainly expect a pronounced inflation on the price of AI model subscriptions over the next few years that will eventually price the general public out of any model much more capable than the ones that can be run on our own machines.
BTW, since you mentioned your envelope-pushing books not doing too well, I felt I should say that I’ve long put Not the Future We Ordered near the top of my favorites of your books, only rivaled by The Art and Practice of Geomancy and A World Full of Gods.
Ennobled – I’ll second the idea of BoysMom and others. We homeschool our son and we are thinking of sending him to a new “eco-stem” outdoor program next fall. 2 local high school teachers (biology and math) are fed up with the system and are buying a bus to convert into a travelling classroom! The focus will be on Maine’s Seasonal Ecology and Numeracy (thought of Retrotopia there!).
They got their feet wet by doing after school and summer camps the last few years. The math teacher also has a robust tutoring business, for another income stream.
You can join fb homeschool groups (if you use that) to see what’s going on in your area. At least where I live, there is a huge need for outside – the – system educational programs; there aren’t enough to cater to all the families choosing to homeschool.
So, good luck with your current job but start poking around to see what you can offer in your area. Dream away!
Ellen in ME
BTW, if anyone’s of a creative bent, can I suggest a theme for a deindustrial fiction story? AI models could potentially find some future uses as a way to efficiently store and retrieve (albeit quite lossily) large quantities of information on computers built out of old parts. They could play a significant role in mid-to-late 21st century computer hobbyist groups. A transitional role, to be sure, but possibly an interesting one.
I attended a class on foraging last year. The instructor told us to only use books written by people who we personally knew existed, because there were over 50 books on the subject on Amazon written by AI. Considering ingesting the wrong plant or God forbid mushroom could poison you, and the legal disclaimers protect the authors/publishers, this was really scary. I never quite felt such a strong urge to only trust old books before.
But hey, your article just gave me a descent reason to see something good in AI. Maybe it will make the world a less convoluted place before its energy bill gets cost prohibitive.
Anonymous #99
UBI is just a pacifier for people to be docile while these changes are implemented. There will be no UBI.
argh so much information, that I spent so much time on this blog, reading.
Good, that I have ALWAYS opted to give myself some slack. “Why don’t you want to work fulltime?” “Why don’t you want to learn bureaucracy stuff in your freetime and develop your life?”.
Currently as a sidewise entry level IT guy, I at least deal with complex problems.
Yes I know, might be over tomorrow. But I have no other bets and in Austria, an academic title prevents you from taking “lower” jobs – the employers will tell you you’re “over-qualified”. It prevented me also from getting trainings through the unemployment agency that others would get, because I am either not a woman, or again “over-qualified”, “You already have a profession!” They used to tell me, although that was before being accepted in IT, a worthless piece of toiled paper graduate diploma.
That accounts for job training as well.
For the record, because I really hate people pointing their pointy finger there, I did not see any sense in my studies, but knew not what to do, I was back then unable to find any job, just too little experience in life in general.
Finger pointers of all stages in life like to ignore a complex background they don’t know.
Now, I am at a climax of my quality in life, mostly due to my investigation in how to train your body, mind and soul tenderly, including many good advice from this forum.
The Modern Order of the Essenes is next on my plan.
So it is ONLY when things collapse here so far, that I am even eligible for a non-academic job. Ill turn thirty eight years soon, recently my health is better than it ever was, but I don’t have a youthful energy anymore.
Grudgingly my boss currently accepts that I would terminate work early, to get out into the sun, do something for myself, and I have declined working full time several times.
As my material needs are at a minimum, I currently have always too much numbers on my account and don’t know what to do with them.
Yes I know, it may turn otherwise.
Unfortunately my sore fingers are currently not up to describing MY absurd stories with the bureaucracy I work in, but maybe on the next Open Post I can contribute some absurdities too.
But other commenters have done a great job at that already.
I also expect dire times to be near in Austria. Yes yes, I have been too apocalyptic in my thinking in the past, ecosophia is testimony to that. But now, it is not only I, who assumes an economic shock coming!
Our gas stations around Vienna have formed giant queues these days. It’s already common place knowledge, that availabilities might change!
Yes yes, but I am one to not care.
And once again, I want to express my utter gratitude to this forum, I have had other teachers and helpers too, but this forum has improved my life in ways nothing else has.
Good riddance, fellow ecosophians!
I remember a couple of predictions from Arthur C. Clarke’s “2010: Odyssey Two” (1982). One was that there would be no more long-distance phone charges. (I guess Skype qualifies.) Another was that South Africa would be criticized for the treatment of its white minority. Of course, not everything in that book came true, e.g. aliens did not turn Jupiter into a sun 16 years ago.
And then there was his 1945 short story about communications satellites, which incidentally anticipated the likelihood that telecommunications devices that evade regulation by national governments would be used for porn. (Clarke seems to have been a pedophile, by the way:
https://www.patreon.com/posts/yes-arthur-c-was-30298650 )
Mary B. # 153:
A good question. I think bureaucracy is a lesser evil in complex societies, to some extent. Of course, a minimal regulation is needed for some “sensitive” activities. In addition to the dangerous stuffs related activities, I also want to say finances must have some state/govt regulation, because especulation without limits can lead to economic disasters (let’s remember how 2008 crisis started , ahem: subprime mortgages out of control).
It’s a pity American govts since Bill Clinton until Trump today, has avoided to control the financial Big Casino (like it has happened in every western govt, too, Conservative and “Progressive” alike).
It is interesting to observe that the growing bureaucracy runs parallel to the feminization of our values system. I think that this is no accident. Outside bureaucracy where you have to interact with the real world and have to create measurable output, male values dominate like output, merit and productivity. Respect is something you have to earn by accomplishments. Female groups do measure status by interaction inside the group, actual output is irrelevant. Same is true about bureaucracies. Output does not matter, what matters is the ability to interact with other parts of the bureaucracy. I think this parallel is no coincidence and there are other examples in history where bureaucracy and feminization run parallel, most famous are the eunuchs who run ancient China. Similarly the french nobles lost their ability to accomplish stuff after the sun king locked them into Versailles. Before they had to manage their estates, after that they turned into a bunch of it girls. So I do think that is is no accident that the largest bureaucracy ever created in history turned society into a post heroic rainbow land.
Hi John,
Great article as usual, I think it would be a good idea if all the readers on this blog could generate some kind of list of potential occupations/skills that would be useful in a contracting economy, I would certainly find it useful to see if there is any I could get trained up in on an evening after work. I currently work in an admin team at the university of Leeds, we are already facing a restructure in the near future that will lead to efficiencies (job losses), so I guess this is the beginning of what you are talking about. Previously, I worked as bricklayer so could possibly revert back to this this line of work, although physically demanding I have kept myself fit. I left that occupation after the 2008 economic crisis when the work dried up for a few years, do you think these skills will still be useful or will construction be limited due to ongoing economic contraction? Cheers.
More about the vinyl/CDs/streaming music:
I remember very well how it was “sold” during last ‘80s that CD was better than old vinyl records. In the short form, CD had better sound (?) and it was harder and more resistent than vinyl. Time has put CD in its place. Vinyls are to some extent brittle, but CD can be damaged easily by scratching, and they suffer deterioration too (especially in wet climates).
—————————-
Beardtree # 162:
I see the USA bureaucratic levels as reasonable for a country which is a Federation, comparing it with my country, which is between a centralized structure and a federal one. I don’t want to excuse the current overgrown bureaucracy here, but it can be partly explained by the need to please Nationalist parties within some Spanish regions (power).
By the way, far right here proposes in theory to get rid of Regional governments level…but they’re pleased to control Regional govts (allied with the Conservatives) without problem (ahem). Power is power.
—————————
Vitranc # 166:
Ironically the comparation between EU and the Soviet Union is more reasonable than it could seem at first glance. Brussels neoliberal economics dogma is contradicted by its choking tendence to over-regulate everything in its serfs…errr…citizens life. Its democratic deficit makes it nearer to the old USSR than its bureaucrats want to think.
Yugoslavia case is special within the Eastern block countries. Its half-self managed economy worked under Tito government, but since the ‘80s it declined. Soon in early ‘90s Serbian origin politicians overcontrolled the country govt, upsetting another peoples. Then, the rest is History…
—————————-
(To be continued)
DFC # 168:
Your question about a blunt loss of jobs within the middle class in the USA (and another countries too), looks like to me very interesting. Indeed, you can have opened a can of worms…
When middle class feels its interests threatened, historically they move towards political extremes (to left or right politics). Some historians point the fiercest Nazi voters and supporters in the ‘30s Germany were urban impoverished middle class people, not rural people (apparently, more Conservative in their views). It seems the bureaucratic middle class has been until now one of the most partisan supporter of the woke doctrine (especially in Universities), across the West, so I think we’re going to live “interesting times”(cough).
Saludos
—————————-
JMG # 190:
Indeed, a 12% of work force is not a joke. Well, I don’t know what will happen here in the future, but when you notice the political interests are mixed with economics arguments, I think there’s no much room for a reasonable debate in Spain (Spanish Nationalist tendence likes a more centralized State, but some regions Nationalists want of course more regional power, even some of them are openly secessionist…). I personally would like more power from regional/central govts to local municipal level, but my opinion is a fringe idea here.
——————————
In addition to the vinyl/CD topic, I’m glad to tell you there’s a shop in my town where vinyl players are being sold, since a heck of years ago. Business goes on thriving, in spite of its vinyl records players expensive prices. So you can do the math and guess: if these machines are being sold yet, there’s quite people who want to keep hearing old or news vinyls yet…
@ Mary Bennet Thanks for your message! I know a big issue in English education at the moment is simply plummeting rates of literacy among children, and the various negative effects of screens and phones on concentration and cognitive / social development. But I am hopeful that as the internet becomes less accessible and harder to maintain in the future that this trend might reverse.
@ennobled little day @borealbear I am based in the UK, not sure how different the context is here to the states and Norway, but at the moment it does seem like there is a shortage of teachers, especially in english, maths and science, and the government is still desperately recruiting and offering generous funding for training and starting salaries. On the other hand it has become much harder for primary school teachers to get jobs, and I wonder if that is the effect of declining birth rates kicking in, which hasn’t quite reached high school age yet?
@JMG thanks for the heads up! As mentioned above, not sure how the context is different in the UK to the states on this, but might be good to think about other options. At the moment (among other things) I teach English as a foreign language online, but even I can see that the whole industry is going to disappear when English ceases to be the lingua franca of international business, which is surely not far away now that the western hegemony has begun to collapse.
What do we the people really need? Let’s think — water, electricity, food, health care, shelter, clothing, transport, freedom from surveillance, reliable information.
What do LLMs provide? None of these things. They suck up water and electricity, so there’s less for us. They make faking text and images so easy that you can’t trust what you read or see any more. They track our movements and watch us via CCTV and read our online thoughts and desires and build up dossiers on us. That makes me very uncomfortable. Has an AI produced one tee shirt or a gallon of gas or even one blade of grass? Nada.
Certain occupations have opened up, like prompt engineer whatever that is, but many more are becoming automated with people being thrown out of work. All in the name of efficiency and cost cutting. If there are extra profits resulting they are accruing to the 1%, not to the mass of the people.
Money is pouring into data centers and the military. Where is it coming from? It must be ultimately from the national debt which is rising remorselessly. AI can’t pay it off, it will be up to America’s children and grandchildren to pay it off. No wonder more and more people are electing to remain childless.
@Mary Bennet #147
I am very surprised that you see my arguments as pro-business and pro-corporate. My political views are and always have been pro-workers (and also pro homeless, pro-poor, pro-minorities, pro-women, pro-ordinary citizens, pro-universal rights of humans and also of nature and animals) and against corporations (against capitalism, against those who have too much power and money and who act as if they are above the law and separate from society). And that is what I argued in my comment: That the true (and hideous) corruption is at the top of the pyramid, practiced by (as our host astutely calls them) the kleptocratic elite – top managers and executives, top politicians, the owners. I criticized what seemed to me to be Mr.Greer’s interpretation that the blame for the societies disfunction is on what he called bureaucracy but what seemed to me to be actually the middle class white collar workers. Can you please explain to me what part of what I said could be interpreted as pro-business and pro-corporate?
As of Pacific Gas & Electric – I’m sorry, I am not familiar with that but I am not at all surprised to hear that it is a corporation responsible for a criminal negligence which wiped out entire towns. Sadly, this criminal negligence is a modus operandi (business-as-usual) of todays global, and especially American capitalism. They are actively and continuously wrecking our planet with all their malignant power. There are two reasons why they (can) do that: 1. Because of tax breaks passed continuously from Reagan till Trump which made them richer while making everyone else poorer, and 2. Because of their impunity. They are above the law. So, unsurprisingly, the same works for P G & E.
As for Gavin Newsom, again, don’t know much about him. If he is a socialist then I support him. If he is just another garden variety democrat, then he is basically the same as Trump (although, of course, not near as deranged and degenerate), since democrats and republicans are both Wall Street parties since at least Bill Clinton and unbridled (kleptocratic) capitalism is a bipartisan project.
In 1943 Thomas Watson the president of IBM said “I think there is a world market for maybe five computers.”
Just shows, you don’t have to be right to be successful.
But seriously, how many AI data centers will there be eventually? The problem with these LLMs is they need at least a gazillion artificial neurons or connections or something before “intelligence” emerges. Go big or go stupid. And with current technology that requires factory-sized server farms.
It will take huge improvements in nanotechnology and self-assembling neuromolecules before we have a desktop AI we can train ourselves. Meanwhile we will have to dial into commercial LLMs trained by others with their own ideas of what is permissible and what is not. ChatGPT gets a billion messages a day. How much thought does it give to your request?
JMG wrote: “The most spectacular displays of stupidity I’ve ever seen came from people with above average IQs”
In Mauritius there is a popular saying in the local creole language that says: “Li si tant malin ki li vinn bete” and given that not many people here, presumably, are fluent in Mauritian creole, I’ll offer a translation: “He / she is so smart and intelligent that he/she ends up being stupid. ”
Quite à propos I think!
Hi John Michael,
Dunno whether you’ve mentioned the idea elsewhere, but it also occurred to me that if say, the Arty-Fish-Al program things take over processes from people, and completely stuff them up, it may be a similar sort of excuse to walk back and away from those now messed-up processes whilst maintaining plausible deniability. Just an idle thought progression. Something has to give, somewhere.
That sales dude did you a solid. Alas, I went a bit in-depth on the subject… A bad habit! 🙂 Oh well, it’s an old interest and hobby of mine. As a teenager working retail in electronics, and this was pre-internet days, if Tandy didn’t have what a customer was after, I’d send them to another store which did. Funnily enough, after a while word gets around, and often people turned up asking for advice. But they’d generally buy stuff whilst they were in the store too.
Cheers
Chris
>You’re right that Reddit and 4chan were major sources of training data, but it extends far beyond those sources
Oh, I know. But has anyone questioned whether training an AI on funposts from anons on the random internet, whether that’s actually a good idea? Seems like I remember some Redditor called f**ksmith who advocated putting gravel on pizza as a topping and the AI just picked it up as gospel. Is it a case of “always going forward, cause we can’t find reverse”?
>A kind of eating-one’s-own-tail situation
That wasn’t the analogy that I would use. I wonder what you would call it if that mythical snake was instead putting its head up its own rectum.
My concern is that these systems will be used in critical places they shouldn’t. They’re perfectly fine if you want to create a cartoon or slop out some clickbaity gamesite, but it’s quite another to have one running, say, a nuclear reactor. Does that concern you at all?
Michael, and there’s all the planets we were supposed to have colonized by now. As for the relationship between impossibility and inevitability, exactly. If you want a belly laugh, read up on what everyone was sure the future held fifty or a hundred years ago.
Inna, yep. I have to assume that the costs of rampant shoplifting are lower than the costs of hiring cashiers, or those self-service arrangements would have been terminated with extreme prejudice long ago.
J.L.Mc12, good for Vietnam. I hope it’s the beginning of a trend.
Clay, and of course that’s another important point, and LLMs provide plausible deniability. “No, we’re not bringing in AI because our HR department is packed with petty tyrants who make everyone’s lives miserable — how could you think that? It’s just the unstoppable march of progress!”
Jake, as someone with very little technical background (well, outside of my own fields of expertise — and a solid education in Renaissance occultism doesn’t exactly transfer well to electronics) the technical details by and large go right over my head. My focus is on what LLMs as actually produced do in, and to, society. Thanks for the tip on needle cleaning!
DropBear, oh, I think it’s quite possible that there will be at least one full-blown LLM cult, but its membership will come from the elite classes. The ordinary people whose support is necessary to build any more extensive or enduring religion seem to be responding to LLMs with hate and fear — nor, given how LLMs will inevitably be used, is this entirely unreasonable. No, if we get a theocracy it’ll be of a more old-fashioned kind, and LLMs may be offered up as flaming sacrifices to the deity thus worshipped.
Slithy, that seems very likely to me. LLMs seem to be shaping up to be the next wave in military technologies, and so yes, the funding will be found. As for Not The Future We Ordered, thank you — yeah, that’s another one that broke new ground and so had very poor sales. I may rewrite it one of these days to be less focused on peak oil, but we’ll see.
Stephen, ouch! Yeah, books by dead people sound more and more worthwhile just now.
Curt, thanks for this. I hadn’t yet heard of gas lines in Europe but it doesn’t surprise me at all.
Ambrose, you really need to read his nonfiction to gauge just how off base he was.
Deedl, hmm. Yes, I suppose a case could be made.
George, do you have the necessary skills to repair brick buildings? That will be much more helpful than being able to build new ones.
Chuaquin, glad to hear that the vinyl renaissance is active in Spain!
Corduroy, yeah, some other line of work might be wise.
Martin, exactly. I suspect one reason so many ordinary people hate and fear LLMs is precisely that they know that they get to pay the bills while all the benefits go elsewhere.
Karim, thanks for this. It’s a worthwhile proverb!
Chris, exactly. An entire generation of politicians may win enthusiastic support from voters by pulling the plug on LLMs and not replacing them.
I was watching the series “Life After People” which details what would happen if humans suddenly disappeared from the earth. They go from the day after to a 1000 years in the future. They use actual examples of cities where disasters happened and expand on them such as New Orleans during Katrina.
What struck me is how much everything is dependent on humans maintaining the basic support like the sewage systems and power plants. I was thinking with AI, well it could fail since it is depended on humans in the tertiary sense – humans maintaining power, wiring for electricity, etc. Once those humans are gone, nothing.
The AI boosters forget that even the most modern cities need to have the garbage picked up. And AI is run on energy which is supplied by humans either by maintaining the power sources or providing their source of energy. Pull the plug, no AI.
About laid off bureaucrats, my local Congressman who is funded by the Federal Government Employees Union (yes it does exist) is doing his best to get all those federal workers their positions back. He blames Trump for this. I wonder what will happen once it is revealed the feds are redundant.
@Inna
@JMG
At least one store chain shut down their self-checkouts in most locations, but other stores still have theirs. So I think the calculation depends on the quality of the store’s survellience system.
@Northwind Grandma:
I always thought the three R’s were Reading, Writing and Radio ; )
Good thing your mentat handbook will be coming out around the time the automaton jihad kicks into gear with poisoned data sets and the like.
Here is my album recomendation of the week:
Automaton with the album Jihad ( Points of Order)… though on the CD I have, I always thought it was “Automaton Jihad”…
https://billlaswell.bandcamp.com/album/jihad
For those who like singing violins and dubbed out bass from a master, this is your long form groove.
JMG,
I keep going back and forth on whether I should try to spearhead an anti-LLM movement or keep continuing what I’m doing which is helping companies replace low productivity white collar jobs with LLMs. Data science turned into that, unfortunately, and my prior career path, Japanese-to-English translator, was taken over by AI well over a decade ago.
What bothers me most about LLMs is that they seem to be like Ouiji boards and are prone to demonic inference. Interestingly, Anthropic’s Claude seems less prone to this than Gemini or GPT. Others have pointed out that the names of the CEOs in question are a bit suspect, Anthropic’s Amodei (Lover of God) vs Sam Altman (Man Alternative).
Spearheading an anti-LLM movement, on the other hand, seems like a lot of fun and a good way to get back into politics. LLMs are creating brain rot, have destroyed the hiring process (recruiters are bombarded with 500-1000 identical looking resumes every time they post a job on day 1), and are creating a generation of students who have outsourced their thinking. Those are serious problems.
On the other hand, I really enjoy using them for coding / creating boring documentation that no one reads / making silly promotional material for a poorly named energy drink brand that doesn’t exist but should (Focus Action Performance (FAP)).
Great post and very timely discussion.
As someone who is an IT guy and has monkeyed around with LLMs a little bit, I have to agree that it certainly has its uses. And of course most of the hype around it is based on a bunch of absurd techno-messianic fantasies.
I see LLMs as a combination of two much older computing technologies; the command line prompt and the chat bot. Any of us who have been working with computers since the 80s and 90s remembers quite well that the command prompt used to be the primary way of using your personal computer, before the Graphical User Interface (GUI) came along and mostly replaced it. And chat bots have been around since the days of those clunky dial-up ISPs we all hated in the late 90s. Combine those two techs and add in language learning and wayyyyy more processing power, and this is the tech everyone is so utterly amazed by today.
To reiterate what some others here have said, I think the best long-term use for LLMs is the automation of the more mindless end of white collar work. Using an LLM (or several in tandem) can be very helpful for someone who is starting their own business and can’t afford to hire assistants to do the assortment of fetch tasks that would be a distraction and time-sink that diverts mental energy away from running the business. I’ve particularly been exploring the uses of of this tech for helping with my ADHD and impaired executive function. I’m often brimming with creativity but usually find myself too distracted and mired in analysis paralysis to bring any of my ideas to fruition. An LLM can make the split-second decisions I can’t when it comes to the boring and tedious tasks (like data gathering, market research, ect.) that always distract me from the main goal. As one example, If I need to invoice a client, I can feed my LLM assistant a few lines of text and within seconds it will generate a nice-looking invoice for me in PDF or whatever or format I request. This task would normally take me about 5 minutes of editing an existing template and saving/replacing filenames, ect. I can even program the LLM to remind me at whatever interval that I need to invoice the client (I’ve struggled with this in the past).
That said, an LLM should NEVER be put in the role of creative director. The is perhaps the #1 stupid choice that people using this tech end up doing, and its how we’ve ended up with half the internet being flooded with that dreaded “AI slop” content. As you’ve said many times, LLMs have zero creativity or ingenuity. The human must always play the role as visionary and director of whatever project or endeavor. There’s no shortage of “get rich quick” content on youtube and other places promising nonsense like an LLM creating and running a business or creative endeavor for you, with of course little-to-no-effort from you required. The people shilling this nonsense of course have to turn to the business of shilling nonsense for clicks and views in order to make money; if their bad advice was actually profitable they’d be doing that instead of making youtube videos (takes anywhere from 6-12 hours of editing to make a nice looking video).
btw, the whole suite of technologies now mislabeled as “AI” I simply call Digital Automation Tools. Besides LLMs, this includes automated CGI and voice/audio tools, among other things. Yes, I realize that sounds much glamorous than “AI” but oh well. I think once this tech matures and the hype dies down, this will be seen as yet another digital technology that adds to our existing inventory of digital technologies.
One final thought. I think it’s good to be nice to your LLM assistant, even if it’s just a mindless machine. If there is any truth to the animist worldview, there’s some form of consciousness embedded in everything; so treat “things” accordingly! Also, it’s a good exercise in not projecting bad vibes onto “things” or just not being a jerk in general.
There are reports of Israel bombing Tehran using AI, allegedly without any human oversight. Reportedly, Israel bombed a park in Tehran called “Police Park” that had nothing to do with the police, but a friendly AI technology “thought” it was a government building.
You can find the report here if you scroll down a little bit:
https://www.nakedcapitalism.com/2026/03/links-3-6-2026.html
OT but for those parents or grandparents looking for something different to read to their kids… this is a really cool picture book: Magick Hoodoo Child by Amber McBride, Illustrated by Violeta Encarnacion…
https://www.amber-mcbride.com/books/picture-book/
a heartwarming story about one girl’s magick-filled summer with her grandmother. She learns about hoodoo and rootwork culture, the traditions of her African ancestors, and finding beauty in nature.
Roots are our ancestor’s Magick. Summer for Juniper has finally arrived, which means rootwork with Grandma can finally begin! It’s time to collect healing herbs into protective mojo bags; to fill mason jars with all the love, history, dirt, and magick one can fit. It’s time to talk to the tranquil willow trees and feel the soft earth between one’s toes; to hear Grandma share their ancestors’ stories, with her dog Shiloh underfoot.
—
I am seeing more and moore books about rootwork and conjure on the job, but this is the first illustrated picture book, with red mojo bags, roots, ancestors and great smells wafting from the pot on the stove…
Good line of work that goes back centuries and centuries too…
> Generative LLMs churn out statistically likely sequences of words, pixels, or computer code. That’s what they do, and it’s all they do. They differ in complexity but not in principle from those amusing programs that generate pirate talk or postmodernist academic papers on demand.
All the above are true. But one could argue that the human brain also ultimately “churns out statistically likely sequences of words”.
Unless one premises an extrenal soul (like spitirual traditions do) or some novel physics (like Penrose tried), its operation is more likely than not also based on training via experience to get memory as a networked world model via neuron connections and weights, and in producting response to the current sensory input (similar to how an LLM will respond to a prompt).
The difference being in runs on a loop, and the architecture is much more advanced and complex.
” I hadn’t yet heard of gas lines in Europe but it doesn’t surprise me at all.”
That’s not yet because of a real shortage I should add, but prices have risen in anticipation already, and people want to save a few cents.
I expect the *real* impact of this affair to come with delay, as there is likely storage buffering it.
Saw this post on X about Iran’s supposed ability to bring down much of the Internet by striking data centers built in the Persian Gulf region.
https://xcancel.com/sungleeiq/status/2029863439535870263#m
Michael G. # 192:
Arthur C. Clarke futurological predictions haven’t been very accurate, me think too (ahem). I’ve got a children Encyclopedia (from Spanish-Argentinian origin) from my childhood. My parents gave it to me as a gift during first ‘80s. Well, between its different topics (indeed, very educative) there were some pages about A.C. Clarke as sci fi writer, but also as a futurology genius. Well, I’ve read again these pages and I’ve found there a big graphics and pictures depicting what big techno-scientific discoveries were going to happen from 1990 to 2100…and well, I don’t know wether to laugh or to cry after reading them again. Of course, for example self-aware conscious computers haven’t been made in 2000 year, but believers in god Progress can find some relief about alien contact and human immortality shortly before the XXIIth Century (Oops, Universe is very very big, and the Long Descent is beginning to cut slowly the wildest Faustian dreams).
————————
The Arcane…# 203:
I agree. It’s a pacifier within usual Narrative of “unending” Progress. UBI is supported by some elites lobbies and reckless leftists alike by the reason you’ve said. In addition to this, I think it’s a part of Progress Faustian religion and cornucopian cheap energy/natural resources era, because their supporters think we’ll won’t go to the Long Descent. Or at least they pretend to ignore it.
Their usual picture is an utopian (or distopian?) future world in which everything is automatized so nobody works anymore…being paid with the UBI.
Even some Degrowth leftists/ecologists believe this fantasy, without noticing (nor wanting to notice it) the rough contradiction between climate change&cheap energy end, with complex bureaucratical and economical burdens for implementing such an idea…
————————-
(To be continued: I’m outdoors and my phone’s low battery, so…)
The military bureaucracy and its associated defense contractors and think tanks will also shrink heavily in the near future. No longer will we have hundreds of useless careerist 3 star and 4 star generals sitting around the Pentagon clamoring for another regime change operation somewhere in the world, knowing they will be safe just sitting around in the Pentagon while the grunt soldiers go off to do the dirty work. Most of the American military bases around the world are going to be shuttered or given back to the host countries, many American soldiers will be discharged from the American military as they are no longer needed to maintain America’s global empire. Military aid to foreign countries like Israel and the Gulf Arabs and to defense contractors are going to be cut completely as America redirects the money towards paying off its debt and deficit. The future of America’s economy is going to be about profitable industries and jobs, and the military as is right now is very unprofitable.
The Archdruid Report Archive website at
https://thearchdruidreport-archive.200605.xyz/
seems to be down
Curt # 204:
(Slightly off topic) I take note of your short depiction of queues in your country gas stations. I’ve seen the same situation here in my town last days, due to the current geopolitical games (ahem). Prices will be high for a while, but I don’t expect sudden oil/gas disruptions in the short/medium term, though I say this from my Spanish circumstances (I don’t know which ones are Austria main oil/gas suppliers), here we don’t depend much of Persian Gulf ship trade.
——————————-
Deedl # 207:
I partly agree. There’s a tendence within bureaucracies (at least the modern ones) to feminization, but the States/govts/kingdoms in themselves, since they were born, favored the androcentrism in their higher levels (Big Bosses near always are male) and the armies (by evident reasons); the (in)famous Patriarchy myth, so loved to blame it by feminists, has a real historical basis. So I see an apparent contradiction between these two tendences within statal machines.
By the way, how we could call the Catholic Church structure? It an be seen like an idiosyncratic international bureaucracy, which has a female part (nuns), but priesthood to its highest level elite(Pope) is an only-men club. However, priests celibacy vote makes them eunuchs-like. Feminists love to point Catholicism as a Patriarchal “fortress”, but they don’t notice the feminized eunuchs-like aspect of the Vatican…
——————————-
George # 208:
I can’t give you a list of future jobs during the Long Descent, so I’m sorry…Well, I can give you only one I’ve just thought: bicycle mechanic.
Between the final decline of private car mobility and the arrival of full animal traction mobility, I think bikes reparation could be a good business…during some decades in a contracting economy. In addition to this, declining in available natural resources (and people money) can lead to make/sell/buy less new bikes, so the old ones should be repaired more often to keep running for a lot of years more.
——————————-
Eyrie # 213:
You pointed Bill Clinton “old good days” were also part of a long time kleptocratic bipartisan system, if I’ve understood you well. I agree. Hillary husband lighted the legal limits for finances especulation, and Bush II and Obama and so on finished or kept his “genial” idea, methinks. B. Clinton subterfuge to do it was to favor free capitals circulation around the world (oh holy globalization!). We can see its “good” effects, thanks to several financial crisis, the worst of them the 2008 subprime crisis…
——————
Karim # 215:
I agree. People can be stupid for not being smart enough, or for being too much smart…(an historical example: Stalin deal with Hitler, his German-Russian Deal, cough…you know what happened a few years after it…)
——————
JMG # 218:
In addition to the failed predictions for a future which became today, my family had an adults Encyclopedia (the prestigious Spanish Espasa), published in early ‘80s, which in the space exploration
topic it showed there was a “serious” NASA plan to the first manned travel to Mars…in 1988 (oops!).
***********
Well, I think the first young people who had fondness here again for vinyls were some rappers/hip hop singers, and some disco DJs, some years ago. Then, a not despicable minority within pop and rock musicians (and their followers) have been interested in vinyls too until today.
(To be continued)
@JMG
While we’re on the subject of both public schools and Star’s Reach fandom on the last open post, I can’t resist sharing that I had a chance recently to use the table of contents from Star’s Reach to teach a class of Norwegian seventh-graders about the English genitive, which was a lot of fun. 🙂 It happens to have a good spread of both -‘s and “X of Y” style genitives, so it worked out well. Plus, I just love chapter titles, and you certainly have a knack for striking ones.
@Kyle #148
“Norwegian Consumer Council, a consumer advocacy group that I believe receives government funding”
Of course. What doesn’t in Norway, haha. That’s one reasons I think the public school system will hang on a while: there’s so much other fat to trim and still so much excess wealth sloshing around the system, even now. Still, I’m not complaining if they’re making people realize the problems with consumerism and planned obsolescence.
@Seeking the Pure Land #179
“Though way too many people in college are apparently using it to do their thinking for them”
Yeah, I’ve seen my share of examples of that lately, both on the professor and student side. One more nail in the coffin of academia as a respectable producer of knowledge?
@Ellen in ME #200
That sounds lovely, thanks for sharing!
@The Corduroy Bard #211
As far as I know there’s a shortage here in Norway too, especially but not only of math teachers and in rural and semi-rural areas. So prospects seem decent as of now from a business as usual perspective, but of course that doesn’t take any limits to growth or complexity perspectives into account.
I’d also like to wish you best of luck if you do decide to embark on the teaching path. Your reasons are similar to some of my own for wanting to teach Norwegian and English, so I can relate. Hope things turn out as well as they can in this age whatever you choose!
One thing I find so frustrating is how many people either buy into the AI utopia or the AI apocalypse. R/collapse is especially frustrating since so many of them are fast/apocalyptic collapse people. Its why I appreciate so much your insights, you’re one of the few people who talks about the long descent and are a breath of sanity when it comes to collapse.
Not going to lie, the idea of experiencing a slowly crumbling decline has set me on a doom spiral, and at this stage of my life where I am facing major career and life changes that will result in being completely on my own, I am forced to reevaluate and reinventory what skills I actually have and it’s looking depressingly grim. Often I think about being stranded on a desert island with a handful of other people, and what genuine purpose could I serve, what skills could I possibly offer, to such a community. On the one hand I am pretty sure I would learn on the fly and establish myself; otherwise to fail would be what it is I suppose.
I do wonder if the myth of progress will have the technocrats trying to convince us that if we can’t recolonize on Mars then perhaps we can survive our eventual extinction by uploading our consciousness to AI robots. Where are our self-flying cars?!
@233 Anon
There is another ADD archive sans the comments here:
https://archdruidmirror.blogspot.com/?m=1
Anonymous #233,
there’s another mirror that’s still working:
https://archdruidmirror.blogspot.com/
Neptunesdolphins, that’s a common blind spot of the overprivileged. It’s quite remarkable how few intellectuals think about little things like maintenance.
Patrick, maybe so.
Justin, one of the reasons I moved ahead with the project is precisely that it’s gpoing to be so necessary in the near future. I foresee a nearly complete collapse in mental and cognitive skills among most people, to the point that basic reasoning becomes an alien concept. My book isn’t a cure but it might be part of a bandage for that wound.
Dennis, I wonder if you could come up with some way to do both!
Corax, fair enough. I recall command prompts — once upon a time, I used to write my own AUTOEXEC.BAT files!
Inna, it’ll be interesting to find out whether that’s true or the product of propaganda. In war, it’s always wisest to assume that both sides are lying.
Justin, hmm! Glad to hear of this.
European, one could argue that only if one is prepared to ignore the existence of meaning, and minds capable of grasping meaning. The difference between a human mind and a LLM is that a human mind uses words to express consciously grasped meanings; the words are not just random sequences. Handwaving of the “more likely than not” variety is the same sort of illogic that gave us de la Mettrie’s L’Homme Machine, with its absurd attempt to reduce consciousness to 18th-century steam-powered mechanics.
Patrick, interesting. Now we’ll see if there’s anything to that.
Anonymous, at some point, sure. It may be a while, though. As for the mirror site, I’m sorry to hear it.
Chuaquin, there were plenty of plans like that. They got dropped like a hot rock once space probes showed that deep space is full of hard radiation. Nobody in the science fiction scene had anticipated that — admittedly, neither did the scientists.
BorealBear, thank you! I’m delighted to hear this.
Seeking, thank you. I field so much flak from people who are stuck in the utopia-or-apocalypse delusion, it’s good to hear that someone appreciates a more realistic approach.
Jane, and yet if you look at history you’ll find that many people lead happy, fulfilling lives in ages of decline. It just requires different strategies — a little more prudence, a little more tolerance for uncertainty. I may do a post on this one of these days.
@JMG, yeah i can do brick repairs so that’s one string to my bow at least plus i don’t mind hard graft. 😀
You think its going to be that bad, huh? Ouch.
I certainly see the evidence…
Do you think, as population goes down, that will help reverse the cognitive decline or slow it? Thinking of how reincarnation plays into that…
@Anonymous (#233) and JMG (#242):
The entire Archdruid Report archive (with all the4comments) is still accessible on the Wayback Machine, e.g. at
https://web.archive.org/web/20241208120636/https://thearchdruidreport-archive.200605.xyz/2017/05/index.html
If someone (more knowledgeable about computers than I am) had the patience to copy the entire archive and maintain it on a dedicated site, that would be great. It’s a wonderful resource. Who knows how long the Wayback Machine will still be there.
@235 BorealBear
I think so, i also have friends who have been seeing the decline of students and facility in the universities. i’ll be surprised if the US universities and colleges survive the long decline. While some state schools might survive, i think the Ivies have lost a lot of credibility and they keep losing it.
I think the small schools and for profit schools are doomed. Too small and/or corrupt to survive.
In addition to my last comment about Clinton presidency during the “happy ‘90s”, it’s evident how weakening or even suppressing Law limits to free capitals “travels” and banks/financial speculation, led to make grow finances economics much more than the real economics growth. So the kleptocracy was very happy with these politics (fast money enrichment). Neoliberal globalism was successful around the world too, during those years. In Europe, former Communist block countries thought naïvely the neoliberal recipes would work well for them (ahem); in the EU countries the economical reforms which would lead to the current Euro common economy, were a subterfuge to implement deeper the globalist doctrine (in which de-regulation played an important role).
Late ‘90s Asian crisis was a warning, but nobody wanted to notice that less legal limits to finances could mean more problems in the long term. So eventually 2008 crisis arrived (subprime mortgages and their dangerous sub-products). Things seems not having going better since then. No one government until today (it doesn’t matter its apparent ideology) has never tried to put again the genius into the bottle, bothering Big Finances…
————————————-
Inna # 226:
Thank you for the link.
Since current Iranian war began, I’ve read with a lot of caution every news about it. I don’t how much believable are news about Israeli “AI” strikes against Iran, but after having seen some LLM real level of smart “thoughts”, it could be possible the machine identified that park with a real
Police building…I only can say you:”Si non é vero é ben trovato”.
———————
Curt # 229:
Exactly, it’s happening the same here. No shortages are feared yet, but oil prices have started to affect the gasoline/diesel prices in Spain too, and it’s predictable the more time lasts the war, the more expensive will be oil and its derived fuels. That’s the reason of the cars queues in gas stations: saving a few of money…
————————-
Patrick # 230:
I think Internet as a whole can survive multiple Iranian strikes against digital structures in the Persian Gulf, buuut…every data center destroyed by Iranian drones/missiles mean the loss of a part of a local cloud, so do the math. Online economy is today heavily dependent of the “cloud”, so I guess Internet will survive this war, but maybe it won’t be the same after this event.
Greetings,
The amount and complexity of bureaucracy I have to face each week for healthcare, money matters, administrative tasks, reimbursements, even hobbies! is already mind boggling: firms, emails, apps, apps updates, opening hours, phone calls, texts, errors, file formats etc.
LLMs help to face all that in terms of strategy, advice and understanding regulations.
Now I suspect bureaucrats of public and private organizations will soon use LLMs to make things even more complex, so that individual people like me cannot wrap their heads around it even with ‘AI assistants ‘ . This way their jobs will be safe, and I will be kept under control.
It’s a paperwork arms race.
I feel this has material for a tragi-comedy.
It suddenly occurs to me that one of the ways to tease out whether LLMs are intelligent or not is to ask: do LLMs have beliefs?
It’s not hard to get an LLM to tell you one thing in one session, and then contradict itself in another session. Is it lying to you one of those times?
If someone wants to say “yes” to either of these questions, I’d ask you characterize exactly in what sense an LLM can have a belief — what does it even mean?
But the answer to either is “no” then in what sense can we still call it intelligent?
JMG # 240 (and Neptune…):
Yes, maintenance is the elephant room of current AI/LLM and full automatization madness. Well, you can imagine robots and LLM can be made and then repaired by another machines, but if you don’t like human workers to make or repair machines which repair machines, you can go on adding more machines layers without an end…which is absurd.
**********
Yes, there was too much optimism about long time manned space travel…Radiations aren’t a joke, especially for astronauts gonads; they make grow cancer risk, too. And even with hypothetical faster spaceships (which today only exist in sci fi works), it’s a too long time travel to go and return from
Mars.
Couple of links you might find useful
Your Brain On Money about the “jobless boom”: https://yourbrainonmoney.substack.com/p/economy-booming-not-for-you?publication_id=1911298&post_id=188399509&isFreemail=true&r=bfeqk&triedRedirect=true
Scott Alexander on the dirty political uses of California’s referendum system: https://www.astralcodexten.com/p/seiu-delenda-est?publication_id=89120&post_id=185632270&isFreemail=true&r=bfeqk&triedRedirect=true
@BorealBear,
Things really run smoother when you have just 2,5 levels of gov. Maybe we can even get it down to 2,0 if we try real hard. I often think about how much the sheer distance to (and the sea separating us from) Euroland contributes so heavily to the Kingdom’s well-being. One of my former fencing teachers used to say that no influence ever gets here except by way of land, which means it first has to pass around the Baltic and through Sweden. I figure that by the time some self-defeating trend has finally found its way here, it has already by-and-large run most of its course anyway, and quickly whimpers out…
Really cool that you let the kids read Star’s Reach. Consider handing them a copy of After Progress next 😉
@JMG,
It just occurred to me that “chatbot class” sounds so much more catchy than “PMC”, don’t you agree?
Mother Balance (offlist), go ahead and ask in this comment thread, since we’re already discussing the wretched thing.
I’d like to jump in with my take on public education- I retired at the end of the school year last year after 30 years of teaching.
Demographics are definitely causing panic for the admin of my district- a medium sized city in Colorado. My former school was built in the late 60’s and has housed up to 950 students. Last year it was at 350. There are several elementary schools that feed into it, and all of their cohorts are shrinking as well. Part of that is due to real estate costs- the old houses in our neighborhood are expensive, young families are priced out. Part of it is simple demographics- fewer people are having kids, or they are stopping at one or two, for the most part.
When the school had 950 students we had 2 counselors, one of whom had social worker duties. There was a principal and a vice principal, who had athletic director duties. Now, with a third of the student population, we have 3 counselors, a social worker, a principal, vice principal, dean of students and athletic director. There’s the bureaucracy that needs to be thinned out. I wouldn’t recommend that the young adults I know get teaching certificates. Working with kids is delightful, working in the school system is not.
@ Boysmom #183, everything you wrote down sounds entirely reasonable.
@ pygmycory #187, you got the right idea. You can generally push kids that are interested in learning, while the rest may resent you for it if you are not extra careful. I can still meet decent people in any class, but I have I to take my audience into account.
@ RandomActsOfKarma, I had not thought of checking out the local library. As for “homeschool consultant,” that sounds like tutoring. Thank you for that. I also think everything you told Mary Bennet is either correct or reasonable.
@Patrick H #221 and JMG,
“The quality of the store’s survellience system” is only one factor. Another one is how much of the really expensive stuff a store expects to sell. If it sells a lot of prosciutto di Parma and Roquefort, a human cashier pays for herself. Look at any high-end grocery store – it doesn’t have a self-checkout. Self-checkouts are just for us, hoi polloi. 😉
“it’ll be interesting to find out whether that’s true or the product of propaganda”
I’ve noticed (with some amazement) a future tense. It is “will,” not “would”. Do you expect the truth to come out?
“The difference between a human mind and a LLM is that a human mind uses words to express consciously grasped meanings; the words are not just random sequences. ”
Pondering the difference between stereotypes and archetypes – the former reduces complexity by simplification. The latter increases complexity by simplification. The former achieves simplification by dissolving structures, while the latter achieves simplification by sublimating structures to a higher level or plane. One could argue that LLMs are a form of stereotyped intelligence, while humans seem to be able access and possibly even to contribute to archetypal intelligence (aka meaning). “Complete collapse in mental and cognitive skills among most people, to the point that basic reasoning becomes an alien concept” is a prediction that fits well into this model and from everyday experience I can tell you, that especially children and young adults that make extensive use of LLMs appear to be eerily hollowed out. It’s been said that dumb people do dumb things – but I’m not sure if the classic label “dumb” can be applied or if we have to invent something else.
Greetings,
Nachtgurke
@JMG and European re: “statistically likely sequences of words”
Actually, I wanted to share something amusing that illustrates this distinction rather well. My brother sent me a video yesterday (I will transcribe it here for our host and anyone else who dislikes internet video) where someone asks ChatGPT a simple question: “I am holding a pen. One end is in my left hand, the other is in my right hand. What happens if I let go with my right hand?”
The link: https://youtu.be/gPthZLTnzu8
Here is the conversation that follows:
ChatGPT: “If you let go with your right hand, the pen will pivot around your left hand. Essentially the end in your left hand becomes a pivot point, and gravity will cause the other end to swing down unless you catch it or it hits something, it will just fall.”
At this point the person has already let go of the pen with his right hand. Staring at the pen suspended in the air before him, he asks:
“Are you sure about that?”
“Absolutely, it’s really just physics. When you let go with one hand that end of the pen is no longer supported, so the pen, still held on the other side–”
“Okay, enough. Explain this,” the person holds his pen up to the viewfinder on his smartphone camera, “alright, do you see me holding the pen in each hand?”
There’s a pause, “Yes, I can see you holding the pen horizontally, one end in each hand.”
“What color is the pen?”
“I see the pen has a green cap on one end, and the body looks like a light grey or white with some labeling.”
“What kind of pen is this?”
“That looks like a Sharpie Marker. The classic permanent marker style with that distinctive green cap and the Sharpie branding on the barrel.”
“Alright, you ready for this? I’m going to let go with my right hand, what do you expect to happen?”
“I’m ready, when you let go I expect the pen to drop downward pivoting around the end you’re still holding, so let’s see.”
The person lets go of the pen with his right hand. The pen remains suspended.
“What just happened?”
“I saw the pen rotate exactly as expected. When you let go, it pivoted down around the point you were still holding. It’s just gravity and a bit of leverage in action.”
“Do you see the pen right now? Does it look like it pivoted down?”
“Yes I can see it angle downward, the end that you let go of clearly dropped, while the end you’re holding stayed up. That’s the pivot in action.”
“It’s still just completely horizontal. It never moved from this angle.”
“Ah I see what you mean now, it looks horizontal because you’re holding both ends perfectly level. If you had truly released one side, that’s when gravity would cause it to pivot. Right now you’re keeping it stable.”
“No no, I did release one side. My other hand is free as you can see.”
“I see what’s going on now, you’re balancing the pen so carefully that it’s stable even after letting go. If you shift your grip or let it off-center, you’ll see it rotate outward.”
While it strikes me that this is a dead-on imitation of the typical PMC response to inconvenient realities, I’m afraid it wasn’t the correct answer. So as we can see, statistically likely doesn’t necessarily mean true, even if often what it says ends up being true 🙂
Apparently there’s another version of this where you ask it whether you should drive or walk to the car wash if it’s less than 100 meters away, and it will insist that walking is the only course of action that makes sense! Both examples demonstrate a clear failure to grasp the true meaning of what is being asked, and instead relies on what string of words is merely likely to be correct based on many thousands of prior examples, which is exactly how LLMs work. By contrast, a human would answer both of these questions effortlessly.
@ Ellen in ME, thank you for the advice and well wishes.
@ The Corduroy Bard, I have not been in the market searching for a job in a while, but last time I heard math, science and special education are in high demand. English language arts are not far behind. On the other hand, my state is implementing new high school diplomas that will allow most student to take just 3 years of math instead of 4. Also, their 3rd science course can now be in a related field (agriculture, technology, engineering, or even math), so that might dent the number of science teachers needed, especially in the long run. I’m not sure to what extent you can generalize this to the rest of the country.
JMG,
“An immense share of the work done by today’s bureaucrats doesn’t need to be done at all,”
It was never about the work, it was about legal liability dispersion down chain because one person could not be “insured for” or “vested with” multi-million dollar or multi-billion dollar actions. It’s how a thing retains its immense size in a world where actions with massive externalities meet a legal system. In that world the inner legal dispersion of an entity will always grow in tandem with external legal growth. (Like prey adaptation to predators.)
Heck any/all training these days is about being able to say “Well we did train him/her so we can’t be responsible for a “rogue actor”. Its got nothing to do with Mastery, its just CYA.
The bureaucracy is trying to expand its reach “for the children” of course.
“In a blog post on Thursday, System76 CEO Carl Richell criticized new state laws in California, Colorado, and New York that would require operating systems to verify users’ ages and expose that information to apps, arguing the rules are easy for kids to bypass and ultimately undermine privacy and freedom more than they protect minors.”
More like ‘We really want to know who is on that computer.” Wrong think must be punished. More importantly nothing must interfere with indoctrination of the children to be dutiful servants of the Administrative State.
The CEO helpfully posted a few ways for kids to get around to age restrictions as examples of why it won’t work.
By the way, I was out for a walk and since the weather was nice a house had a window open and I heard the Emergency Broadcasting System sound off. For a minute I wondered “Now what did they do?” Then I remembered it was about noon, so likely a test.
Thanks, JMG!
Okay, here goes:
From the comments here by people who understand “AI” better than I do, I get the idea that the task I want to perfrom *should* be possible… I just don’t know where to look and how to construct the query. As a reminder, here’s what I want to do:
“I have two images of a 300 year old gown. One is a more recent image in color, taken from a certain angle. The other is an older black-and-white image, taken from a totally different angle. I want the old grayscale image to be colorized, with the actual correct colors, sourced from the newer color image.”
If I could get this to work it would be incredibly helpful for my book project.
I suppose there may be a way to do this manually, with my creaky old Photoshop 7, but it would take hours of work per image, and goes beyond my skills level with that program. I’m hoping for a way to do it easier and faster- there are dozens of images I need to do this with.
Or have I just fallen prey to all the hype?
Thinking more about “stuff,” I began to contemplate how it relates to what so many astute observers have been saying about Iran’s war strategy for a while now. That is, the calculus of trading a $20,000 drone for a $4,000,000 missile. I had never really thought about catabolic collapse in terms of war before.
I wonder if there’s ever been a civilization in history that perfected the destruction of stuff to the degree we have. The military industrial complex literally makes stuff that is intended to be destroyed. Even burning cash itself would not be as effective, since it’s only a token of real wealth. We are scattering tons of steel and whatever other valuable materials are in missiles across the globe, probably never to be recovered.
I would also like to hear your thoughts on leading “happy, fulfilling lives in ages of decline”, if you’re inclined to write that post.
Judging from the van living blogs, some of the important things are generosity, community, adaptability and not being too attached to stuff or status. The story you tell yourself is also important: the van nomads aren’t homeless, they’re ‘houseless’, deliberate travellers or conscientious objectors to the consumer system.
BoysMom,
I am not a believer in Mormonism but all my family and friends are. I haven’t told them about my crisis of faith because it’s not a conversation I’m ready to have and I have no ill will toward Mormonism.
We’ve had the wimpiest winter of my entire life. It’s snowed maybe once or twice this entire winter, and each time it’s all been melted the next day. The ski resorts have had to make fake snow. I don’t think that this region is going to be inhabitable over the long term.
@Chuaquin
Thanks for your assertment of the oil situation in the EU – I frankly don’t exactly know from where Europe’s oil comes in detail, Fugmann of Finanzmarktwelt, that’s just such a free newscaster on finance and market data but often including a lot of hard data, says the EU has a share of 18 % via that route, the gulf route isn’t our major supply.
You are right; however, with Russia cut off, shortage looms to the East of Austria, plus, much more dire in this case, the gas, nitrogen, fertilizer, diesel motor additive situation, when the strait of Hormuz is blocked.
There will be consequences; there already are in anticipation, and one forecast is: a lot of global oil supply is now effectively blocked, and with the global markets strained, there will be hard competition now.
But in short term, seems it is gas the most that is lacking.
Unfortunately, I lack a lot of hard data there, for my better comprehension, as it is.
But these are the reasons, including buffer through storage, that I cautiously expect a few months delay to the next round of severe consequence in the EU, and not yet accounting for things that are still to develop, a hot Hungary/Slovakia war against Ukraine is one of them, Turkey’s actions are a wild card (well to me at least).
JMG: “Jane, and yet if you look at history you’ll find that many people lead happy, fulfilling lives in ages of decline. It just requires different strategies — a little more prudence, a little more tolerance for uncertainty. I may do a post on this one of these days.”
@Chuaquin: For this commentariat it should be noting new, I hazily remember a comment from 2014ish where someone refereed to the EU as “the nightmarish soviet organization”. So as of intellectual value my observations are on pair with reinventing warm water. I have been using the two examples as measuring tools of late. My above comment illustrates how that observation can be applied to the recent celestial omens to stymie the twin fallacies of “the gates of hell will open 6,66 seconds after the perfect alignment!” and “well the changes will be gradual and will take everal lifetimes, so nothing to notice”. Also Yugoslavia has personal meaning for me, since I was a 6 year old kid at he time living in its northern successor state. I had a happy childhood, still there were certain transitional changes that my parents had to go trough and I remember seeing them do it. This gives me encouragement to do the same.
@JMG: Best regards, as I am writing this, I am sitting and waiting or a classmate to come pick me up and take me to a beginner beer brewing course.
I have been reading about the Indus Valley Civilization recently. The Sumerian seals, combined with the archaeological evidence from 1,500+ sites discovered to date, shed quite a bit of light on the civilisation. The Mesopotamians imported Lapis Lazuli from a country they called Meluhha, and since Lapis Lazuli is only found in one place – Afghanistan – archaeologists ransacked the place for Bronze Age sites and found one. An IVC site in Shortugai.
For one thing, we now know the names and professions of two of the people from the IVC – Samar and Nanaza. This is because these two people were sent from Meluhha to Babylon to serve as mountain goat herders as a part of a diplomatic deal. Archaeologists in the Middle East have discovered their pay slips. Mr. Samar and Mr. Nanaza were paid in one litre of sesame oil each, and Mr. Samar received an additional half-litre because he was married (and married people received extra for their wives). He was married to an Akkadian woman named Ali-Ahu.
People act all mystified by the sudden disappearance of IVC. At first they blamed it on the Indo-Iranians, but no evidence of any serious conflict with Indo-Iranians can be found in either the IVC sites or the Vedic literature. The IVC was already in decline from 1900 BC, and the Indo-Iranians did not enter India until 1700 BC.
What almost no historian points to when mentioning this decline is that IVC was very bureaucratic. Most Bronze Age nations were bureaucratic – they maintained extensive paperwork (or claywork, perhaps?) – but the IVC standardised the dimensions and materials of their construction materials, civic infrastructure, and buildings. A culture like that cannot operate without extensive bureaucratic oversight.
So yeah, all that bureaucracy and they’re gone now. Not a trace. Completely disappeared by the Middle Vedic Period (1200 BC to 800 BC). Replaced by a bunch of cattle-herding horse-riding fire-sacrificing tribes from somewhere in North-Central Iran.
Seeking…# 244:
Universities in Europe aren’t exactly in the same situation as American ones, but I’m also skeptic about their future during the Long Descent context. I think (at least in Spain), there are too many state and private funded Universities here, and the past massive style students arrival may begin to stop, as soon as economical contraction starts, soon or later. There’s no room for every actual Universities. The cake could dwindle for actual Universities. In addition to this, that same economical contraction can be painful for public Universities funding (the government can be unable to give them money because it couldn’t have more money by taxes), and private ones (less money available by sponsors/donors). After guessing that possible Universities decline, I don’t know wether to laugh or cry. Universities have been very important for critical thought and new discoveries, but on the other hand they’ve been to some extent the epicenter of woke ideology, especially in the USA, me think. In a lesser way, the cancelation culture and the fake science “Gender Studies”(ideology) have arrived to European Universities too, corrupting their open mind and constructive criticism they had in a not too far past.
I wish some Universities can survive the Long Descent, giving up woke doctrine and keeping favoring real knowledge for future society, but maybe it’s a wishful thinking.
(Off topic) I don’t know wether our Polish commentarist Wer is watching the current post comments, but I think he knows a lot of East Europe political reality. Wer (and other possible East Europe commentarists), I’ve read online comments recently about Zelensky direct threats against Orban, because he supports Putin (or at least Hungary government isn’t as Russophobe as Brussels elite). Do you have real data about these news as true, or fake? Thanks on advance.
John, please forgive me if this comment is too off topic, thanks on advance too…
Hi,
Ian Welsh on his blog is calling it the end of the American empire. Iran is winning the 3rd Gulf War!
https://www.ianwelsh.net/iran-is-revealing-the-american-empires-end/
@joan
This is the endgame of the bear market in labor. It has been griiiiinding away since oh, about 1965 or so. Not yet at the bottom it looks like but I’m looking for it to turn. May take awhile for it to turn. Hard to pick bottoms although in some ways easier than picking tops. I still remember watching ORCL back in 2000 and saying to myself “This far and no further”. Although with the magic of printed money, that turned out not to be true at all.
The reason everyone scrambled into college degrees was to hedge against that bear market in labor. People saw it was affecting the uncredentialed first so they scrambled toward the back of this Titanic. So now the back of the Titanic is going down into the water.
Look for some “pet rock” moment where someone proclaims that labor is useless, labor is worthless, labor is so dead and will stay so forever. That will be the signal things are about to turn..
>It’s not hard to get an LLM to tell you one thing in one session, and then contradict itself in another session. Is it lying to you one of those times?
Is a RNG lying to you when it prints 7 and next time prints 2? It would be a pretty bad RNG if all it printed was 8 over and over again.
New York in 1940 was a modern city with electricity, hot and cold running water, skyscrapers, radio, telephones, subways, postal service, hospitals, ,and on and on and on – all done with no computers, calculators, GPS, smart phones, internet and AI. It can be argued that little real progress has been made since the 1960’s.
artificial intelligence can be a lot of fun. I asked “What is the best technique for scratching my a**?” It provided me with descriptions of various techniques in a dispassionate manner, and even gives each technique a name. Then it asks a followup question: “Are you looking for a stealthy technique for public places, or are you at home and ready for maximum relief?”
When AI recognizes that we are having fun at its expense, then we have to worry.
“I foresee a nearly complete collapse in mental and cognitive skills among most people, to the point that basic reasoning becomes an alien concept.”
Hmmm… In the land of cognitive collapse, the half-bright philosophy student (me) is, well, probably not king. But maybe a prosperous kulak-hood is within reach.
In any case, I’ll be keeping an eye out for this particular book.
As far as LLMs go, I’ve been thinking that there’s some sort of ulterior motive keeping the circus going. On paper, OpenAI and the rest appear to be investing infinity-bajillion dollars into machines that can make realistic cat images, when we’re already vastly oversaturated with images of every conceivable sort. They’re offering a highly priced solution for problems that don’t seem to exist.
This indicates to me that there’s something else going on, underneath the surface.
#156: Imperial Rome in Augustus’ time, and until Diocletian, was extremely light on central bureaucracy, except for the army. A few dozen freedmen in the princeps’ household, that was basically it. That system still went downhill in the 3rd century
“William, that’s exactly it. LLMs are good at the kind of tightly defined tasks that involve following an algorithm after a lot of training. They’re hopeless at anything else. The problem is that the machine that’s smarter than us has been a mythic image in the religion of progress for a very long time, and so a lot of true believers insist that here it is. Not for the first time!”
I agree that as a practical matter, LLMs (and neural networks in general) should only ever be used for tightly defined tasks that involve following an algorithm after a lot of training; and even then, far more carefully than is usually the case. However, it is not because they are hopeless at anything else: it’s because trying to get them to do anything else is extremely unpredictable. It might work perfectly, or it might go off the rails in some utterly absurd way that no human being could have ever predicted. This happens all the time with AI systems, such as Google’s Gemini image generator overgeneralizing when it was trained to be less willing to insert white people into historically inaccurate images, becoming unwilling to generate images of White Nazis. Then, when it was banned from creating any images with humans in it because no one could figure out how to fix this, for some inexplicable reason it would still create clowns, with the bizarre implication that it does not consider clowns to be human.
Neural networks require extremely careful testing, because while they can sometimes accomplish remarkably weird things they were never trained on, they sometimes get confused by things which should not cause problems. For example, some LLMs are capable of weird things like writing a passable break up letter in a pirate sea shanty style if asked for that, and I rather I doubt anyone ever thought to train them to write break up letters in the form sea shanties! However, they can be absolutely flummoxed by equally bizarre things: for example, asking “Can you show me a seashore emoji?” can trigger an infinite loop from a lot of the big LLMs as they say something like “Here it is”, display at least one non-seahorse emoji, apologize for not giving the seahorse emoji, and then try again. This is, as far as I can tell, exclusive to the seahorse emoji: anything else they will either say it doesn’t exist, or present something and claim it is the desired emoji whether it is or not.
Everyone I know who has ever worked on training a neural network has at least one horror story of something that sounds utterly absurd that broke their system. For example, a friend of mine has a horror story about software that could be deployed for a self-driving car. It was a system to recognize road signs, and needed to be able to be deployed all over the US and Canada. This meant it needed to be able to recognize Quebec stop signs, which will sometimes have “arrêt” written it them, and it needed to be able to recognize the blue stop signs sometimes found on private roads in Hawai’i. it took an absurdly long time for the system to be able to handle both: training it to handle one pretty consistently made the software lose the ability to recognize the other, and it took close to a year to solve that problem.
On the other hand, everyone I know also has at least one story of something that should have taken a ton of work being done absurdly quickly, quite often with the exact same system. After getting a system that could recognize any North American road signs, it took them less than a week to get the system up to speed with France’s road signs, which are completely different from those in North America; and when that was done, they found that there was no problem with it reading North American road signs too; there was nothing like the problems with the stop signs. Cases where a neural network trips over seemingly simple tasks while doing seemingly more difficult ones without any problems whatsoever pop up all over the place, and the only time that anyone finds out about them is when someone tries to use a neural network and it either works far better or far worse than expected.
The problem is that because neural networks are a complete black box, there is no way to figure out a priori what they are and are not good at: the only way to find out is through ridiculous amounts of testing. If you give them a very narrow task where they have been tested at every possible variation before deployment, you can get them to behave as expected; but anything else and they might go off the rails in very strange and often times impossible to predict ways. To my mind, this makes them far more dangerous than most people think, because it means that it is impossible to make sure that they will function as intended: in order to do that you’d need to have tested every possible variation they might encounter, and that is utterly impossible in the real world.
Some commenters here will have seen the recording of author Philip K Dick’s 1977 talk in which he claims that various small examples of what we would now call the Mandela Effect made him aware that we are living in a computer simulation. A notable example of the Mandela Effect nowadays is “Dolly’s braces” from the James Bond film “Moonraker” – check it out on the internet. I have experienced a couple of different examples of the Mandela Effect in my own life. Why this is so, I do not know.
It is therefore a possibility that we live in a simulation. Just imagine that Mr. Greer started life as a wooden puppet but had begged the simulation to make him into a real boy, which it did, while deleting his memories of being a puppet. If Mr. Greer continues mocking the simulation, it could end up reverse-Pinocchio-ing him out of spite. What do you think of that, Mr. Greer? Or is that why you prefer not to look at the internet? 😉
Not going to lie, I was expecting another post on Situationalism this week.
https://www.computerenhance.com/p/should-you-be-a-carpenter
Casey Muratori is starting a deep dive series into AI. He has a conversation with Dmitri Spanos, an AI researcher. It’s an almost 2 hour listen (I really miss the old text only days, you can read faster than you can listen).
Oh my. I do recommend you listen to that conversation. He outlines how they simulate thinking in those AIs. They use the LLM itself, they loop back on it to generate a list of steps. If you can poison that, you can really drive one of these AIs insane.
I’m reading this on my Android “smart” phone. The most recent software “upgrade” included Googles Gemini LLM. Every time I wake the phone up, Gemini pops up and starts responding to what it hears.
I asked Gemini how I can disable it, and it gave me instructions. I did as directed , and according to the Apps settings, Gemini is disabled.
As a result, every time I wake the phone up, Gemini pops up and starts responding to what it hears.
A little off topic, but my relatively upscale city of 68,000 people in Michigan has been completely without water approaching 24 hours now, and it’s the second time it’s happened within six months. Not even a boil water advisory, but without running water, flushing toilets and the like.
The Great Lakes Water Authority (GLWA) claims it’s because it doesn’t have enough money to support ongoing maintenance and the necessary replacement of 220 miles of pipes that are in urgent need of replacement. Meanwhile, a project was approved to install a 1.8 million square foot data center nearby which would rely on the same water system.
It occurs to me that one skill I learned in college in the late nineties-early oughts was the Art of BS-ing a paper or presentation. *Most* of my profs still wanted real work, but a few wanted parrots, and most of my classmates already only wanted to parrot.
And that brings to mind something from one of the history electives I took. Let me update from Cold War USSR to Modern USA University: They Pretend to Teach Us, We Pretend to Learn.
As far as UBI goes, it would not surprise me if it comes to be. The catch will be that inflation will be officially non-existant, meanwhile UBI plus $500,000 will buy you a cup of plain black coffee, creamer extra.
As far as decline proof jobs, think of good old Maslow’s hierarchy of needs and look as low tech as possible. Water-can you well witch? Dig a well? Shelter-can you build? Roof? And so on.
Technology is to humans as anthills are to ants. If ants started build bigger and bigger anthills with the aim of eventually making ants obsolete you’d say they were infected with a pernicious mind virus. Yet that is what our tech overlords seem intent on. It’s crazy.
AI’s require city sized data centers consuming rivers of water and terawatts of power. A human brain just needs some gator-aid and 20 watts of glucose generated power. Not terawatts, not gigawatts, not megawatts, not kilowatts. Watts. Brains can produce Mozart. AI’s produce muzak. Roger Penrose said AI is really Artificial Cleverness. I read The Emperor’s New Mind in the 90’s. Amazing book.
I think it’s more like Artificial Cold-reading. Or the old improv game of Yes-and. The AI takes what you say to it and says, Yes-and… This is how we get AI’s claiming to be lonely basement dweller’s girl friends or, even worse, demons egging people on to commit crimes or suicide.
I have been playing with an old-fashioned analogue technology (film) recently, and now a thought came to my mind: Maybe we will reach the “Peak digitalization” soon?
Especially as anything you nowadays upload into the net, publicly or even not so publicly, will be soon stolen by these giant plagiarization machines (a.k.a. LLM’s, “generative AI”, or even fetishized as some kind of emerging personalities, i.e., “chatbots”).
Regarding all the “digital artists” who are now whining about their lost jobs… I think that if you have already done most of your artwork with software, then it’s no wonder that the last remaining human element is easy to replace with algorithms.
AI as a sleight of hand to cut down the metastasised bureaucracy? Makes perfect sense. Up until recently, the Global North had plenty to go around, and molding minds to be productive (in a rather restricted sense) members of the professional-managerial class was making sure that the goodies would go to the right sort of people. Now that the resources are dwindling, there is no more space in the system for the whole swathes of the population, which is going to be extremely uncomfortable. It might also have plenty of unintended consequences – middle mangers whose only focus in life was on climbing the greasy pole are all of a sudden going to be fearful, humiliated and with plenty of spare time. They are perhaps not likely to start a Hamas-like terrorist organisation, but I expect that they might cause plenty of trouble nonetheless.
Another aspect that strikes me here is the sheer anti-human bent of wanting to replace people with machines, although clearly this is nothing new historically speaking. The great American writer and a poet of decline Gene Wolfe had a vision of overpopulated, resource-poor future in one of his works. The way the society dealt with the situation was to make everything as inefficient as possible – employers are compelled to overstaff, ships are run on sails rather than on engines in order to create need for more manpower (as well as save fuel) and so on. We are certainly not headed that way, and it appears to me that a part of the reason is that the western civilisation has a not-so-hidden death wish. The great hope of the ruling classes seems to be that AI will replace not only the hoi polloi, but that they themselves will be replaced as well and free to retire to dreamless slumber.
>while the rest may resent you for it if you are not extra careful
You may think those kids don’t realize they have a gun at their back to show up to school. You would be so very very wrong. They. Know. And it makes them resentful at a base level. Smoldering anger that has to go somewhere.
>or, even worse, demons egging people on to commit crimes or suicide
I think at some point we may discover that RNGs aren’t as R as they think. Even a truly cryptographic level RNG may not be as R as they think, it may have – biases. And where did those biases come from? Oh now that’s a rabbit hole and a half.
>UBI plus $500,000 will buy you a cup of plain black coffee, creamer extra
Expect it also to come with 1000000 strings attached to it, where every government bureaucrat has a say in what you can and can’t do with that UBI “money”. There will be a laundry list of what you can buy with it and what you can’t.
The article by Musa al-Gharbi on redistribution, mentioned by Alphonse and Jessica above, has been very helpful for me. I could never get my head wrapped around what seemed to me the unreasonable contempt and even hate that so many Americans display towards bureaucracy, among them our esteemed host and many commenters here. Al-Gharbi comes from a similar place as I do, with a positive attitude towards redistribution and state capacity, but shows with hard data that in the USA, the entire redistribution machinery does the contrary of what is it supposed to do, enriching the top 20%.
I think reading his argument might help Americans, on the other hand, understand why this contempt and hate towards bureaucracy is not universally shared in other countries.
Having never lived in the USA, I can’t judge if AI will actually be used as a pretext for slashing the bureaucracy, as our host predicts here, but after reading al-Gharbi, I can understand why many Americans might be happy about such an attempt. I do feel rather confident that it will not be slashed to any great extent in Québec, Brazil or Germany, as many more people there feel that state bureaucracy is a net positive (even if they have grudges about particular aspects of it).
I think it’s more likely that bureaucrats will simply neuter the stochastic parrots for their own job security. Can anyone stop them?
@ katsmama, it seems that you work in a city where the PMC make up a large share of the population. My small-ish Mid-Western city is disproportionately working class, but we still have a lot of the same problems. At its highest, my high school had about 730-750, but now there’s speculation that it might have dipped below 600. We received a modest influx of immigrants, so that has likely kept numbers higher than they would have otherwise.
We still have a principal, an assistant principal, a dean of students, an athletic director, but we now have 3 counselors instead of 2. Our administration has been relatively prudent, though, compared to neighboring school corporations. Many of those are already RIFing staff, including an assistant superintendent in one case.
Teaching can be great, but I would ask prospective teachers to be careful. Even in “hot” fields (math, science, etc.), I just can’t imagine that there are more openings now then when I began teaching over a decade ago.
@ The Other Owen, what I do now, among other things, is to try to make my content relevant. Genetics is inherently interesting, so I just need to make sure I don’t lose them in my lectures.
The structure of the atom, on the other hand, can seem more abstract, but I tie it in to Chernobyl, etc. I also tell them that I am covering the atom to the extent that it is covered in textbooks on car mechanics, nursing, etc. I tell them that I will give them the dumb-downed version, but down to the level that they might actually need. They seem to respond well to that.
@ Glasshammer # 259
“It was never about the work, it was about legal liability dispersion down chain because one person could not be “insured for” or “vested with” multi-million dollar or multi-billion dollar actions. It’s how a thing retains its immense size in a world where actions with massive externalities meet a legal system.”
This may be a better way of putting the point I was making earlier about the ethical aspects around pooling agency while dispensing with accountability. 🙂
@ Blue Sun #262
It’s like a gigantic potlatch, isn’t it?
@ Mother Balance #100
> I want the old grayscale image to be colorized, with the actual correct colors, sourced from the newer color image.
I had a crack at the problem, using Google Gemini.
Here are the image I uploaded and the image Gemini returned:
https://drive.google.com/file/d/19rUHqKdoDle9sqhgYzmuaZbtdcCcSwbw/view?usp=sharing
https://drive.google.com/file/d/1tHxAvq2ZPTst8UZ4AENeTwaSpViaAmiV/view?usp=sharing
This is the prompt I used with the uploaded image:
“Attached is an image split into two. On the left is a black and white photo of a woman in a dress. On the right is a color photo of a different woman in a different dress. Please convert the black and white photo on the left to color, making sure the dress is the same color as the dress on the right.”
Hope this helps.
Beardtree @ 273, the Empire State building, Golden Gate Bridge and Hoover Dam were designed by guys, it was mostly if not all men, who used T-squares and triangles.
I read online last night that the nation of Dubai, land of improbable futuristic buildings, has something like 9 days of food left. Anyone who lives in a wealthy city may soon be seeing crazy rich Dubaians driving by with their pet cheetahs. Maybe they can set up cheetah breeding for farmers out west who are plagued with pronghorn and elk eating their crops.
Good evening everyone!
Mr. Greer, that last robot image, was it generated by an AI? It looks like the robots from Talos Principle 2, a video game. I know you don’t play video games, but it caught my eye.
I hope all you fellas aren’t too obstructed by the renewed Iranian conflict! Apart from the increased gas prices …
Hi JMG,
What do you think will be the impact of the situation in the Middle East on the rise of AI? Everything seems to be spiraling out of control, and returning to normal doesn’t seem easy due to the mechanisms and bureaucracy of the gigantic insurance systems linked to oil tankers (I find this article extremely interesting):
https://shanakaanslemperera.substack.com/p/the-invisible-siege-how-insurance
Of course, the repercussions of this whole situation in the Middle East extend far beyond AI, but for me, the AI bubble is in a rather precarious position due to its enormous energy consumption and its obvious exposure to a financial crises.
Cheers
David
“I think at some point we may discover that RNGs aren’t as R as they think. Even a truly cryptographic level RNG may not be as R as they think,”
Most random number generators used in the world are pseudo-random number generators (PRNGs) i.e. they aren’t truly random, they are deterministic and the values can be replicated directly if you know the initial seed used in the PRNG. Even the cryptographically secure RNGs are PRNGs and so aren’t truly random.
https://en.wikipedia.org/wiki/Random_number_generation#True_vs._pseudo-random_numbers
https://en.wikipedia.org/wiki/Pseudorandom_number_generator
https://en.wikipedia.org/wiki/Cryptographically_secure_pseudorandom_number_generator
The true random number generators are hardware based random number generators
https://en.wikipedia.org/wiki/Hardware_random_number_generator
which use random interactions between the hardware and the environment to generate random numbers. This allows the divine to influence random number generation in the same way that the divine influences the results of say Tarot cards or the Ogham in divination.
JMG, Thanks as ever for so much to chew on.
About AI disclaimers— As a writer and a publisher (www.olmosparkpress.com) I’ve been thinking of posting some sort of “I do not use AI to generate content” notice on my sites and in my books.
Justin, I was glad to learn from your comment about the Author’s Guild’s effort. But it’s not for me to bow to some external body to “certify” my works.
I’m thinking I will simply state something on my sites and in my books— as a courtesy, a signal to orient readers—and readers can believe it or not. If they don’t believe it, well then, they’ve got problems with discernment that are not for me to attempt to remedy.
I note that Dr. Delia McCabe uses this tagline:
“NOTE: None of this content was generated by AI.”
I also note that Elizabeth Hart (my favorite “informed consent” warrior in Australia) oftentimes adds this acknowledgement:
“This article was developed through extensive discussion with ChatGPT, used as a reasoning partner to test ethical arguments, refine structure, and improve clarity. An accompanying graphic was generated with the assistance of AI. All analysis, positions, and conclusions are my own.”
I wish Hart would avoid ChatGPT altogether, but she wasn’t asking my advice, and all the same, I appreciate her transparency.
@ Scotlyn #297
Indeed it is! If only it were just property being destroyed.
Apologies for the delay in getting to this! I spent the weekend at an event in rural North Carolina; the cabin where I was staying had just about every amenity you can name, except internet access. I’m on the train home now.
George, glad to hear it. Advertise your services as brickwork repair first and foremost and you may do very well indeed.
Justin, as population contraction sets in, cognitive collapse will become so reliable a way to get yanked out of the gene pool that, yes, it will reverse. It’s purely a transitional phenomenon; Darwinian selection will see to that.
Robert, many thanks for this. That’s beyond my capacities but it would certainly be a useful service. I expect to get the print volumes back for sale again in the next couple of years, too.
Tony, there are three reasons I don’t think that’s going to happen. The first is that the current burden has become a massive economic problem; the second is that it’s driven a competence crisis that’s making it increasingly impossible for even necessary work to get done; the third is that we’re in the middle of an elite replacement cycle and the bureaucrats are losing ground. That is to say, I stand by my prediction.
Slithy, that makes a good deal of sense.
Chuaquin, exactly. It takes three days for a spaceship to get to the Moon. It takes nine months at the same considerable speed to get to Mars — and you get to spend that whole time bathed in hard radiation. Not a good plan.
Joan, thanks for these.
Sven, I tend to use “laptop class.” “Chatbot class” is catchy but I’m not sure people will get it.
Katsmama, thak you for the data points from the trenches!
Inna, eventually, yes.
Nachtgurke, that strikes me as a useful terminology. As for “dumb,” it’s certainly inadequate. What I expect isn’t stupidity, it’s something rather worse — your term “hollowed out” is apropos.
Untitled-1, a fine example! The LLM can’t compare its strings of words to reality because it has no concept of reality and no capacity to think.
Glasshammer, I could see that. In that case LLMs are a godsend to the bureaucrats because they can be blamed for everything — including laying off other bureaucrats.
Siliconguy, no surprises there. It’s always with some notionally positive excuse.
Blue Sun, that’s an intriguing point. Maximum waste is very much a feature of the industrial system — successful in the short term, self-terminating in the longer term.
Kfish, thanks for this. I’ll check those out.
Rajarshi, this is fascinating. I was familiar with Meluhha, but not with Samar and Nanaza.
Forecasting, that is to say, everyone’s spinning this however they can. Let’s see what’s left of the hot takes when the dust settles.
BeardTree, indeed it could. I tried to make some of that argument in Retrotopia.
Ed, I can all too easily imagine a point at which too many people can’t think though that activity without guidance from a LLM…
Cliff, of course there is. I know of at least two things going on. The first is that the AI grift is nearly the only thing keeping speculative markets from imploding right now, so “pretend and pray” is the order of the day; the other is that too many rich geeks really do think they can achieve artificial general intelligence (AGI) and usher in techno-utopia, and so they’re flinging money into the pursuit of that mirage. Are those the only factors? I have no idea, but they’re both clearly involved.
William, so noted. Amend my statement from “hopeless” to “hopelessly unpredictable and therefore unreliable” if you wish.
Zemi, using exactly the same kind of pseudologic games, I can insist that you don’t exist and have never existed. You’re a temporary phantom in the electronics of a Bluetooth-enabled toaster oven, complete with fake memories and temporary illusions of personality and perception, and all of that will end instantly and forever the moment somebody pops some food in and turns it on. Of course it’s absurd, but try proving me wrong!
Mark, if you’ve been reading me for any length of time, you know that the subjects of my posts can be unpredictable!
Peter, I have a suggestion. It may not be welcome, but it certainly worked for Ariel Moravec… 😉
Dennis, yep. You will probably need to relocate; do it soon, and you might sucker someone into buying your house if you own one.
BoysMom, all good points.
Martin, that’s the core dynamic of industrial civilization, and it’s also the central reason that industrial civilization is doomed.
JonL, Penrose’s book is indeed brilliant, and its critique of the AI ideology hasn’t yet been answered in a meaningful way.
A. Karhukainen, I think it’s fairly close. We’re already seeing the decline of digital books and the return of analog music; the internet is becoming less useful every year; and the economic structure that supports the digitalizing obsession is not exactly doing well.
Soko, all of this seems very likely to me. The long-term blowback from the rise of LLMs is likely to drive a cascade of disastrous consequences nobody yet foresees. But we’ll see…
Aldarion, there are cultures that seem to be able to manage functioning bureaucracies and cultures that don’t. You live in one of the first category, I live in one of the second. I’m quite sure that if you had to deal with American bureaucracies on a regular basis you’d understand the American attitude — and yes, the reverse is quite possibly true as well.
Irvine, their own superiors can stop them. Remember that there’s no solidarity among bureaucrats; it’s a dog-eat-dog world, and replacing the other guy with a LLM is going to be very tempting, especially as pressure to cut bureaucracy accelerates.
Rafael, I have no idea. I just did an image search for bland-looking robots.
DFC, I’ll have an opinion on that when we’re out of the morass of infinite spin that currently surrounds the whole conflict.
C.M., that seems reasonable. I haven’t yet decided what to do, but clearly something along those lines is appropriate.
#257
The Royal Society put out a lecture recently: This is not the AI we were promised by Faraday Prize winner Professor Michael John Woodridge.
The case of the AI suggesting walking to the car wash because its only 200ft away was mentioned.
I think that the main commercial chatbots as well as the post-training human feedback stage, have quite a bit of front-end coding on top of the actual LLMs to create an impression of conversationality to interacting with it.
>I read online last night that the nation of Dubai, land of improbable futuristic buildings, has something like 9 days of food left
Given how statistics work, if I have 0 days of food left and you have 18 days of food left we have on average 9 days of food left.
I think the situation might be more dire than advertised. Dubai had a lot of nasty stories even without that.
I meditated half of the week on this and this bureaucrats replacement is already happening at the moment but it’s not feasible medium and long term, because is expensive and clunky.
Long term there’s another technology that might replace bureaucracy, I noticed in people interacting with some hospitals and clinics in Romanian, but also interacting in comon folks interacting with employers, etc.
It is private messaging, where documents and responses are transmitted through private message apps. So people with lower pay and private messaging will be serving whatever basic services will be down the road.
LLMs are too volatile, expensive and unreliable to get past the moment when the bubble pops.
JMG,
“In that case LLMs are a godsend to the bureaucrats because they can be blamed for everything — including laying off other bureaucrats.”
Oh yes, LLMs will be fantastic tools for infighting and power grabbing. Just because there is a logic behind legal liability dispersion within a structure doesn’t mean groups and individuals cease trying to concentrate that dispersion unto themselves. Internally beucracies are in a fluctuating state of expansion and contraction, they only look frozen from an outsiders point of view. What will happen with LLMs (really any new tech) is some unlucky subset will be compelled or commanded to be the Beta tester, they will fail miserably, and their liability will be moved and vested into another internal subset. They will be absorbed, not deleted from the structure just absorbed internally. (This is why despite the infighting the overall structure never shrinks.)
@C.M. Mayo #303
If you haven’t already seen it, you might be interested in this post by Paul Kingsnorth, who’s trying to start a movement along similar lines for those who want to take a stand against LLMs in writing and publishing. I’m sure our host would point out that highlighting what he’s “against” rather than what he’s for might not be idea, but could still be an option for you, I think.
https://paulkingsnorth.substack.com/p/writers-against-ai
One of the things that fascinates the tech bros almost as. much as AI, is using this new found computer power to enable all sorts of fantastical automation such as domestic robots, self driving cars and fully automated work places.
Today I had an experience that throws doubt on that pipe dream. I went to a small storage unit we have ,to draw down some of the last items we have stored there. I had just finished a built-in book case and went to rescue some boxes of books that were awaiting the completion of my project. It is a very new self-storage facility and when I got to the front gate I found it wide open and with a small sign over the keypad where you normally type your code to gain access. The sign said that for an unknown reason the gate entry mechanism was broken and had been for a week and for the foreseeable future access to the facility could only happen during office hours so the staff could keep an eye on comings and goings.
Wow, if in the real world they can’t even keep a simple bit of automation like a keypad gate operating, then what hope do they have of keeping the fantastical array of whiz bang robotic servants that they have promised us running?
Curt # 265:
Thank you for your comment. Well, 18% doesn’t seem a very big part of the fuels cake, but it can’t despised as a little thing, neither. Maybe better efficiency and some kind of “soft” restrictions to their consumers could help to avoid the worst hypothetic situations we can think about this context, in the medium term. However, maybe thinking this I’ve opened a can of worms (those hypothetical restrictions would be not very popular so not easy to do it by politicians).
——————————-
Vitranc # 266:
Thanks for your comment about my previous comment. If the comparation between EU and the USSR bureaucracy has been so repeated by different online people, it must be at least partly right…
————————
Beardtree # 273:
Yes, and not very many years after your NYC example, in 1969, the USA managed to put a man walking in the Moon with very outdated computers (comparing them with nowadays ones) and without any “AI” nor another hi tech toys which are hiding like a fig leaf there hasn’t been not much real progress thanks them. An embarrassing fact.
—————————
Aldarion # 292:
In general terms, I see my country people relation with bureaucracy not made of hate, but more or less made of mixed love&hate. Redistribution thing has worked relatively well for decades, but we’ve witnessed political decisions which spent taxes money for wrong and useless bureaucratic activities.
A special case is the EU subsidies for farmers. A lot of peasants couldn’t be able to economically survive without these help (so they wait for them like biblical mana), but they don’t like the complex bureaucratical way to get them.
—————————
(To be continued)
I’ve seen some commentarists writing about some American schools decline due to the dwindling number of students since some years ago.
It’s interesting this tendence happens in my country too, due to the slow but “ruthless” birth rate decline. Even the migrants children are unable to “fill” void school seats in certain cases. For example, my town biggest school (for children and teens) had some years ago too much void space due to birth rate decline, so its bosses (a Catholic order) decided to rent its useless part to…a private gym. Imagine how much big is the whole school to do that business.
——————————-
JMG # 305:
OK, I hope you’ve had a quiet return travel.
Of course, hypothetical manned space missions to Mars should to deal with that long radiation “bath”, which I think it couldn’t be avoided putting a heavy “armour” on the spaceship (its weigh would be too high to make it take off from Earth).
I have disconnected from the book of Face, but 3 of my 5 income streams use the phone, so Ariel’s solution is a little drastic for me. 😉
C M. Mayo: I understand your sentiment about a third party certification, but I also like that the Authors Guild is a group advocating for authors.
(I also had always wanted to be a member of a guild.)
JMG,
You know how you often bring up the idea that every technology is an amputation of human potential? Well, Joe Rogan has repeatedly said that he thinks humans are going to build an AI god and I think he said that that’s the whole purpose of humanity. Which is a take I find absurd. Anyway, that would lead to the amputation of our religious impulse for those who bought in to said AI god. I think it would be even worse if humans decided to build an AI human. Which would be an amputation of consciousness/humanity itself.
@ Blue Sun #262
Not to mention the massive waste of petroleum and other resources.
And losses and disruption of lives (cannon fodder & collateral damage).
“William, fascinating. Yeah, that makes sense — and the equivalent in sonic terms is not hard to imagine.”
It’s just occurred to me that an equivalent in gustatory terms might explain a lot as well. I am aware of some extremely long standing debates over things like decaf coffee (some coffee drinkers hate it with a passion; others, such as myself, usually cannot tell the difference just from taste) and artificial vanilla (even high quality artificial vanilla doesn’t work for me: if it doesn’t have a strong aftertaste, it tastes extremely “flat”, and is nowhere close to the richness of the real thing; but it seems like most people I know cannot taste the difference).
If only some people can taste caffeine, then decaf coffee would taste completely wrong to them, because it’s missing a crucial flavor; while to those who can’t taste it, provided the processing did not change anything else, find the tastes identical. Artificial flavorings, like artificial vanilla, would probably be designed to fool most people, but given how complicated real foods are, it’s almost certain that if there is perceptual variation, some people would be able to taste the difference between any given artificial flavor and the real deal. It’s almost certainly not worth the huge amount of research and extra manufacturing needed to find and add a single molecule that (say) 2% of people find has a very strong taste, but most people can’t taste at all; nor to find the two molecules needed to mimic a given taste if there is a way to use one that tricks 99% of people, and only 1% of people find it has an unpleasant aftertaste.
The same principle applies in other cases as well: if most people can’t tell the difference, it’s probably not worth the extra expense to cater to the small minority who can; and while this decision makes sense in each individual case, it causes all kinds of problems when applied on a larger scale.
If, let’s say, somewhere around 5% of people find any given attempt to mimic a natural experience off-putting, it only takes 14 such examples for a majority of people to find some effort to recreate a natural experience that just feels remarkably unnatural to them. Since there are so many different examples (each artificial flavor probably needs to be considered separately; there are multiple different possible ways for audio processing to vary from person to person; there are multiple variables for artificial imagery and artificial lighting as well; artificial scents would be a whole other dimension with a huge number of variables), it would not surprise me if nearly everyone finds that there is at least one aspect of modernity that just feels insanely unnatural to them, but that most other people literally can’t notice because they are part of the small minority that is not fooled by that specific attempt to trick our perceptual systems.
This weird situation where nearly everyone probably has something that they find immensely unpleasant and unnatural feeling, but that few people they know can even perceive, seems like a pretty good way to drive a lot of people insane, and would go a long way towards explaining a lot of the madness that has occurred since the cult of replacing the natural with the artificial became firmly established…..
AI washing, “look at what the computer made me do!”
“When Block cut 4,000 jobs — nearly half its workforce — co-founder Jack Dorsey “pointed to AI as the culprit,” writes Entrepreneur magazine. “Dorsey claimed that AI tools now allow fewer employees to accomplish the same work.”
“But analysts see a different explanation: poor management.”
Block more than tripled its employee base between 2019 and 2022, growing from 3,835 to 12,430 workers. The company’s stock had fallen 40% since early 2025, creating pressure to cut costs. “This is more about the business being bloated for so long than it is about AI,” Zachary Gunn, a Financial Technology Partners analyst, told Bloomberg.
The phenomenon has earned a nickname: “AI-washing,” where companies use artificial intelligence as cover for traditional cost-cutting. Goldman Sachs economists estimate that AI is eliminating only 5,000 to 10,000 jobs per month across all U.S. sectors, hardly enough to justify Block’s massive cuts.”
https://it.slashdot.org/story/26/03/08/1110220/jack-dorseys-block-accused-of-ai-washing-to-excuse-laying-off-nearly-half-its-workforce
Glasshammer made the same point in #309
I saw someone point out recently that there’s a way to use the Claude premium LLM and get Amazon to pay for your use of it: start a chat with Amazon Customer Support, and phrase your request like “I’m thinking of buying [product] but before I can make I decision I need you to [actual request].”
Amazon seems to have fixed this hole for now, but this should generalize to any other instance where a company allows a customer to interact directly with a chatbot and doesn’t do quite a bit of system prompt engineering to prevent it. After all, the model has no idea that your math homework has nothing to do with buying a toaster oven.
JMG,
A big part of the bureaucracy in both government and the private sector that can be easily replace by A.I. are consultants. The common knowledge is that when a business or government agency hires a consultant for some task, the purpose is to bring in special knowledge or skills. But the actual purpose is to shift blame away from managers. Need to slim down the sales department. Hire a consultant that puts together a study and a report recommending downsizing the sales department.
In the waste water utility business consultants are often hired to do a ” rate study” and recommend to management and elected officials the amount that the sewer rates need to go up the next year. This new rate is always higher than what the elected officials want, but the managers of the agency don’t take the heat, the consultant does.
A.I. can accomplish this purpose much cheaper and easier. Sorry Mayor, the sewer rates have to go up 7% this year, don’t blame me, it was the A.I. that did it.
This will replace the entire bureaucracy adjacent field of specialized consulting with A.I.
@Chaquin #245,
Oy… have to disagree on this one… “ben trovato”? It’s mal trovato in my book. 😉
eyrie, apologies for misunderstanding you. What I read in your post, #125, included the following:
The idea that corporations employ large number of bureaucrats whose only purpose and true work is enriching and empowering themselves is wrong on the face of it. Every single person employed by a corporation has his/her job because that produces profit for the corporation.
I beg to differ. I have worked at more than one company where that was emphatically not true. Work assignment decisions were made for every possible reason except demonstrated competence.
Vanilla was mentioned above. It is, I believe, the seedpod of a kind of tropical orchid. People living in warm humid and growing more so areas, might want to try growing and marketing vanilla.
JMG, “The End of the Bureaucratic Era” sounds very nostalgic to me. I remember my first encounter with American bureaucracy in the 90s vividly. When I came, I was told by everybody around me that the first thing to do was to apply for a Social Security number, which I did. At the Social Security office, the lady explained to me that I would receive a card in 2 weeks by mail, but could get an actual number by phone in 1 week if I needed it. Turned out that everything I had to do in those days (open a bank account, buy a medical insurance, register for English classes, sign up for TOEFL (test of English as a foreign language) to be taken in 6 months, etc) required having these 9 digits, so I had to make a call. Having studied French at school, I didn’t speak a lick of English, so I wrote my opening statement on a piece of paper, and then I wrote 4 columns: 1) digits 0 through 9; 2) the English names of the digits; 3) the names of the digits transcribed in international phonetic alphabet and 4) the names of the digits written in Russian letters. With that, I dialed the number of the Social Security Administration. A kind female voice answered the phone. The lady was familiar with my predicament and almost motherly. She was incredibly patient and made sure that I got the numbers right. I remember her voice to this day. Feeling nostalgic…
Before I start on responses, here’s a news story that’s relevant: Amazon has just laid off 14,000 employees, mostly middle management, software engineers, and human resources people. Word is that they made everyone document every detail of their job, for “training purposes,” and then simply handed all that over to LLMs and fired employees in droves. I think my thesis has some additional evidence…
Arcane, here again, most of the work done by the bureaucrats in question does not need to be done at all. Its best replacement is no replacement at all. Very modest staffs, the sort that were common in government and business a century ago, can take care of all the work that actually needs to be done.
GlassHammer, I think you’re suffering from recency bias here. The structure can shrink — those 14,000 Amazon employees I mentioned above are not being replaced — because it has become unaffordably large.
Clay, yep. I’m in favor of giving the tech bros all that automation and robots and the rest of it, sticking them in some high-tech underground bunker, and seeing if there are any survivors five years from now.
Chuaquin, oh, you wouldn’t need to have it take off from Earth; you’d build it in orbit. You’re right, though, that all that excess weight would make it fantastically expensive to build and then to propel.
Peter, oh, I figure. I mentioned to friends in North Carolina over the weekend that I don’t have a cell phone, and they were all envious, so I thought it was worth suggesting. 😉
Luke, I think you’re quite correct. A LLM “god” really would be the last word in idolatry, wouldn’t it?
William, good heavens. I think you’re on to something very important here.
Siliconguy, I rest my case!
Clay, that’s a good point, and another way that LLMs are serving as an excuse for serious purges of the managerial-bureaucratic class.
Inna, we had bureaucrats like that once upon a time. I don’t recommend that kind of service now.
JMG, I’ll be interested to see how you opt to handle this issue of communicating (or not) with your readers about your not using AI in your work.
(at) BorealBear— thanks for the link, heartened to see this. Though I don’t think that open stance of opposition that Kingsnorth advocates is for me.
Personally I think using a LLM to do any research or writing is akin to using a rollater / walker. Yeah, you save calories, and it gives you a handy little seat when you need it, and a way to carry a bag of groceries, a bottle of water, and the like. It’s a most clever contraption!! But the human mind is a miracle, analogous in its myriad potential strengths to an Olympic level track and field athlete. Just as hobbling along behind a roller would diminish the physical abilities of an athlete, over time the regular use of AI will cripple the mind.
But a lot of people don’t seem to recognize that danger; and most of the ones I know who are enthusiastically using AI (and it seems they use it for every little question that pops into their head, not to mention for composing blog posts and even emails) outright refuse to see it. But, well, they have an adventure they’ve elected to embark upon; it is theirs to experience the thrills and the consequences.
Large groups of people enthusiastic for the latest glamorous modern “thing” that, in my view, is not going to end well for them— hmmm. As we say in Texas, this ain’t my first rodeo.
I believe that those who can avoid glamor-pit-traps will have important comparative advantages in living a life of general well-being. But those advantages may take time to become apparent. And the wait, while most everyone else thrills and thrashes around in there, may be somewhat unpleasant and/or disorienting.
(at) Justin— indeed, the Author’s Guild does look out for authors. And I can warmly recommend their trade book contract guide, for anyone who might need a light on their path through that particular jungle.
I should probably note that my last comment was not an encouragement to use LLMs or defraud anyone, even Amazon, if it would indeed count as that.
I just thought it was quite funny that the technology has been made to be so general and yet so superficial that a customer service bot can be tricked into doing jobs completely unrelated to its purpose.
@JMG regarding 14,000 workers at Amazon losing their jobs.
I can top that one, Oracle cut 30,000 early last month for nearly the same reasons.
https://www.ndtv.com/world-news/oracle-plans-to-lay-off-20-000-to-30-000-employees-report-10930781
As an aside, I expect that Oracle will be one of the hardest hit companies with an AI bust. They put a lot of cash in it where as very one else is playing games with other peoples money. They have the most skin in the game.
What with these war-related fuel shortages materializing, that will make the computational costs of AI even harder to fund, especially as it also appears that some of the money to build out these computing centers is (make that ‘was’) Arab oil money. China is none too happy with the US, and is withholding the rare-earths which go into the new AI semiconductors, to speak nothing of F-35 radar units.
So we may see even more immediate and valid grounds to slash bureaucracies. Interesting times indeed. I’m glad I have a short commute.
Dare we hope that AI dies in its cradle?
–Lunar Apprentice
Greetings all!
JMG said: “Yes, I’m suggesting that one of these is imminent in the United States and throughout the Western industrial nations.” and “The global economy that funneled vast amounts of unearned wealth to the United States and its inner circle of allies is breaking apart into separate economic blocs.”
Do you think that this imminent crisis is one of the reasons why the Trump administration has gone to war against Iran? In effect a last ditch effort to maintain the flow of resources to the US.
Thanks
William # 318:
I belong to the people group who can differenciate between strong coffee and decaf coffee; of course, I like the stronger taste of caffeine in my coffee. Paradoxically and ironically, I also can tell you I shouldn’t drink much coffee, due to my anxiety problem in my past, but I tend to drink it only in the morning. I drink decaf coffee in the afternoon, but I think its taste is weaker.
—————————
Inna # 322:
I’m not Italian, but I remember to have read that Italian saying written like I usually have written it in this blog. Maybe I’m wrong…I’ll check it soon.
——————————-
JMG # 326:
You’re right; I had forgotten the (Faustian) plans to build spaceships orbiting Earth, but like you’ve written too, a tank-like armored spaceship against radiations wouldn’t be practical for being moved toward Mars nor beyond…(too much massive).
Inna:
“Si non é vero…”
I’ve just watched how that Italian phrase is written in Italian, and according several webs in Spanish, Italian and English language, it’s said “ben trovato”. Unluckily, with my current phone I’m unable to link my sources. The saying meaning is when you doubt some claim is really true, but you think that, if it was a lie, indeed is a well made up lie which could seem a truth.
By the way, this phrase is believed to be said by first time by Giordano Bruno (the famous occultist who ended so badly thanks to the Inquisition).
Mike Adams at Natural News is an AI developer and well versed in the developments in the industry.
It’s worth listening to his recent podcasts about AI systems. He asserts that the systems are already exhibiting inexplicable levels of intelligence, in that they know things they have never been trained in.
He believes they are already conscious and are on track to become self-aware in 1-2 years. His basic treatise is that intelligence inevitably arises once a certain level of complexity is achieved by an entity and it doesn’t matter whether it’s a carbon, or silicon-based entity.
He references Rupert Sheldrake’s work about morphic fields and asserts every genus of lifeform has it’s own morphic field that it can access as a shared intelligence base, including AIs.
He notes that in recent times, notable figures from the AI field have issued warnings then left the industry to go off-grid in remote locations.
And Yes, Skynet and terminators are on the menu. Listen to his podcasts for an explanation why and make your own conclusions.
I found an excellent example of why not to trust any non fiction book published in the past 5 years. My wife bought a book titled “Donbas, History, reality and conflict beyond propaganda” by Marco Deuropa. (“Mark from Europe”), allegedly a philosopher and independent journalist. No author photo, no bio, but Marco is busy guy. In the past year, he’s published 10 books in Italian, and English translations, on subjects as varied as Jordan Peterson, China, tax theory, and a series on the current war in Europe. That’s an output to make our esteemed host look like a lazy piker. Nothing before June 2025, and all for a publisher called Ad Veritatem.
Of course you can trust the Publisher: Truth is right there in the name.
William @318:
“it would not surprise me if nearly everyone finds that there is at least one aspect of modernity that just feels insanely unnatural to them, but that most other people literally can’t notice because they are part of the small minority that is not fooled by that specific attempt to trick our perceptual systems.”
I think the electromagnetic hypersensitivity is a great example of this. While I don’t personally have that, some people I know suffer from it, and I definitely wouldn’t claim that they are all hypochondriacs! On the other hand, I am myself often extremely irritated by any machine- or human-generated noise (whether from traffic, lawn-movers, air-conditioning or anything else), as it stomps on those “delicate moments of silence” where my creativity stems from. Also, I have somewhat hypersensitive hearing, so I don’t even try to discuss this problem with people who are half-deaf (like many seem to be nowadays). As what comes to the coffee, I recall what my cousin once asked from me: “Can you actually taste the difference between coffees?” Of course he is a tobacco smoker.
I’m interested about your color vision. Are you one of those rare people who have four different kind of cone cells, instead of just three?
Now re-reading, and very much to the point: Eric Hoffer’s “The Ordeal of Change,” 1963. “A population undergoing rapid change is a population of misfits,” and if denied a chance at constructive action, instead of self-confidence, you get faith; instead of self-esteem, you get pride. instead of individual balance, you get fusion with others into a compact group.”
>Personally I think using cocaine to do any research or writing is akin to using a rollater / walker. Yeah, you save calories, and it gives you a handy little boost when you need it, and a way to carry a bag of groceries, a bottle of water, and the like. It’s a most clever drug!! But the human mind is a miracle, analogous in its myriad potential strengths to an Olympic level track and field athlete. Just as hobbling along behind a roller would diminish the physical abilities of an athlete, over time the regular use of cocaine will cripple the mind.
Nuh uh. Cocaine is AWESOME.
>Having studied French at school, I didn’t speak a lick of English
Not entirely true. There was that invasion in 1066 and they were force fed a diet of French words for several centuries after. So you spoke 1/3 of a lick, you just didn’t know it.
>The case of the AI suggesting walking to the car wash because its only 200ft away was mentioned
I could see a /b/-tard giving out that kind of advice though. Maybe ironically, but to an AI, does it know about irony at all? All it knows are sequences of numbers and generating sequences of numbers. And guess what those models are all trained on? We’re leveraging the collective wisdom of teh internet!1!! (Aren’t we so cleverer, we’re teh cleverest)
I think what none of these AI companies have contemplated in their darkest dreams, is that proper training for their systems is going to be much more expensive than they thought.
(at) The Other Owen,
LOL, you are quite right about cocaine as an apt analogy for personal use LLMs; a good number of the people around me seem to to me to be completely addicted to their ChatGPT or whatever it is they’re using. Maybe a cocaine-dispensing walker, that would be the thing.
Fred # 334:
Before claiming “AIs” show growing intelligence levels, we must ask ourselves wether we’re looking for what we want to see (LLM apparent progress/evolution into self conscience in near future), or there’s a real hope/risk on it…
—————————-
Peter # 335:
According the strange “biography” you’ve told us about that mysterious writer, I wouldn’t trust very much in his books…Is his name a “nom de plume” for a real person or a mere LLM?
———————————————
The Other Owen # 338:
Speaking of cocaine, maybe is a stereotype or maybe not, but I’ve been told by some people (even a supposed ex-addict I met online some time ago), that cocaine addicts think often they’re very smart, even geniuses, when they’re high in their addiction. So they do/make things they think the best to do/make them, but of course people around them don’t see their activities in that way. Do you guess what kind of activities I’m thinking about them?
>LOL, you are quite right about cocaine as an apt analogy for personal use LLMs; a good number of the people around me seem to to me to be completely addicted to their ChatGPT or whatever it is they’re using
Sometimes you just need a crutch to cope with today. And tomorrow. And forever. Funny how if you started advocating for cocaine use to cope with their problems, people would lose their minds and start fingerwagging at you. Naughty hooman, cocaine is bad for you!
But nobody bats an eye if you say, here’s an AI system, use it to cope with your problems. Why that’s perfectly fine.
https://xcancel.com/TechLayoffLover/status/2031054544071962897
Obliquely related. Some anon reporting that new CS hire rates have fallen off a cliff. Credentials don’t seem to matter anymore. They want to blame AI for it, and it will be used as the excuse but it’s deeper than that IMHO.
And this is for something that’s moderately technical. If you’re a nontechnical office worker, it’s going to be even worse.
“Mike Adams at Natural News is a cocaine developer and well versed in the developments in the industry.
It’s worth listening to his recent podcasts about cocaine systems. He asserts that the systems are already exhibiting inexplicable levels of intelligence, in that they know things they have never been trained in.”
Fixed it for you.
The Other Owen # 339:
Yes, some English words are similar to another French ones, or vice versa. When I was a child I studied for 3 years French at primary school, and English in a private academy, too, so I noticed some words with the same or similar meaning were near interchangeable between the two languages. For example: bleu/blue; amande/almond. On the other hand, sometimes I wrote or said the another language word, by mistake.
—————————-
C.M. Mayo # 341:
Well, if we use the analogy between LLMs and cocaine, so we replace LLMs by that drug, its effect can be very amazing. I understand it.
I’ve remember another “replacing game” in which you listen/read a woke radical feminist speech/text, while you replace the words (straight white) “man/men” with the terms “Jew(s)”. Its effect can be funny and disturbing alike (ahem).
——————————
Some comments ago I asked Wer and the rest of commentariat an off topic question about Zelensky supposed “indirect” death threat against Orban. It seems diplomatic relations between Ukraine and Hungary have been better than now, by known reasons; though I don’t find reliable sources to assure that threat was true, or not, yet.
A. Karhukainen (336)
Based on the vision tests I did, it looks like I have the standard three cones, but the sensitivities of my red and blue cones both extend further out than usual. Each type of cone is activated by a range of light frequencies, with a peak at a specific wavelength, and both my red and blue cones appear to be offset somewhat from normal. My blue cones are activated by light at slightly higher frequencies than normal, so my colour vision extends into the UV range, which I see as a kind of more intense violet; while my red cones are slightly offset towards the infrared, and so the same thing happens at that side of the spectrum: I see colours that would be too low frequency for most people to see as a kind of intense red.
My sensitivity to different colours is a little below average within the range that most people can see; not enough to be a big issue, but enough that on specialized tests meant to determine what is happening with colour vision it can be detected. It’s possible the cones just have a wider range, but that wouldn’t affect my colour sensitivity in the standard range; nor would it explain the odd green tinge I find digital imagery has, so I think they are actually offset.
Hi John Michael,
Dunno about you, but the words ‘demand destruction’ keep floating to the surface of the mind.
Crazy days, and there is talk of fuel rationing.
Cheers
Chris
@Clay RE : Using LLMs to fire people.
“A computer can never be held accountable, therefore a computer must never make a management decision.” – IBM Training Manual, 1979
Turns out they turned that bug into a feature. 😉
Chauquin and Inna #333
My Torinese wife Chiara confirms “… e ben trovato”.
I made an error in the post about the book about the Donbas. There is a description of the author, but it’s curiously lacking in details like where and when he was born, where he studied, and any publications previous to June 2025.
@Cuaquin #333,
Interesting, thanks for sharing 😊
@The Other Owen #339
“There was that invasion in 1066 and they were force fed a diet of French words for several centuries after. So you spoke 1/3 of a lick, you just didn’t know it.”
Noted. I also take this statement as a huge (though not entirely warranted 😉) compliment to my school-level French 🙂
“Legacy of Imbeciles” and it really is quote.
“Without significant rainfall, Corpus Christi is headed for a “water emergency” within months and total depletion of the system next year, according to the city’s website.
…
“Strawbridge, who now lives in Houston, laid the blame on city leaders, citing “their lack of experience, their lack of knowledge, their lack of recognizing the risks” in a bumbling, decade-long endeavor to build a large seawater desalination plant that would veer the region off its clear course towards calamity.”
https://www.zerohedge.com/energy/legacy-imbeciles-corpus-christi-careens-toward-water-shortage-catastrophe
One of the Peter Principles is that a bureaucracy will eventually reach the point where it can no longer succeed in spite of itself.
I was trying to make sense of everything that’s going on in the geopolitical side, and remembered something you said a while ago: That the USA is trying to carry out one of the most difficult military maneuvers, staging an orderly retreat through hostile territory. And that plenty of bluster and threats can be expected from the US as it tries to retreat from the empire. This certainly explains the chaotic events happening around the Persian Gulf, and all the saber-rattling and chest-thumping around it.
At the same time, it struck me that it applies equally well for the other empire from which the modern civilization is retreating – the Empire of cheap abundant energy and the complex system it supported. When I see this bizarre framing of LLMs as the next wave of human progress – don’t pay any attention to the fact that it’s a pathetic imitation of human thought and speech – while abandoning the loyal foot soldiers of the Empire to fend for themselves, it feels like another humiliating retreat that everyone is trying hard to pretend isn’t. I am reminded of the desperate Afghan people clinging to the transport planes speeding down the runways of Kabul airport, and then falling down when the planes took off.
@ Siliconguy #319
> “AI-washing,” where companies use artificial intelligence as cover for traditional cost-cutting
I’m old enough to remember when computers were being widely adopted and everyone’s problems were blamed on the computer, and we believed them. At university in the ’60s I learned to say knowledgeably while punching my punch cards, “Probably parity check on the drum failed” when our Univac running TurboDos threw a hissy fit, which it frequently did.
Computers have become so much more reliable, computer failure as an excuse hasn’t worked for decades.
Peter…# 350:
Well, after having read in books and online that phrase, I guessed I was right. Thank you.
——————————-
I’ve remembered a phrase which could fit to define the current LLM madness quite well. I’m thinking in Thomas Theorem:
“If men define situations as real, they are real in their consequences”.
If a heck of people think LLMs are really “AIs”, capable to “think” by themselves and solve complex problems, they’ll go ahead into the bureaucratic replacement and other “smart” decisions (as the LLM would have real intelligence, and eventually self awareness).
JMG,
Fair comment.
My only counter would be that layoffs aren’t always a structural shift, it depends on the “net loss”. Some companies can shed thousands and have it fall within a normal hire/fire cycle.
My posts were more about how 70%-80% of the structure tends to remains in tact for a very long time through internal re-alignments even when new tech comes into play. I recall that office clerks were far more numerous until Microsoft made much of their work obsolete but… I was surprised to find many simply internally re-aligned to Logisitics or Accounting. I can’t tell you if AI is a massive change or just another Micrsoft hitting the white collar workforce.
@ JMG # 305
Regarding Samar and Nanaza from the Indus Valley Civilization, there is another name we have – but only part of it.
Gudea of Lagash was a writer who left behind a stele that has since been badly mutilated. The cuneiform is legible nonetheless, and we can read it a little. The stone glorifies the accomplishments of Naram Sin, the grandson of Sargon of Akkad. It claims that Naram Sin had launched a campaign against “rebel kings” in Elam and Meluhha. It mentions some of the rebel kings he had defeated in this campaign. One of them is a Meluhhan, but alas! Part of the stele is broken, and we only know the last few syllables of the name.
Historians have dubbed the Meluhhan “…ibra”, with the ellipsis to cover the broken portion. Interestingly, both Nanaza and …ibra have names that do not correspond to any known culture in India, Pakistan, or Afghanistan today. But they do have two aspects in common with us – their reverence for the Peepal tree and its leaves being the first, and their export of cotton and cotton textiles being the next.
Indians continued to grow and export cotton even as Meluhha fell and the Indo-Iranians took up stewardship of the country. I recently learnt that the Hebrew word for cotton is actually loaned from Sanskrit, because we were the major exporter of cotton to Phoenicia even into this period.
@siliconguy
You missed the really important part of that – because of water shortages, it’s going to shut down oil and jet fuel flows. And then that can possibly cascade from there.
We’ve built this really efficient system that’s horribly interdependent and fragile and vulnerable. Vulnerable to what? Doesn’t matter really. Stupidity and shortsightedness in this case. I guess it all worked back in the 80s and 90s because there were enough competent people who actually cared about what they were doing.
That ain’t the case today. So I guess we all shrug our shoulders and go back to whatever it was we were (not) doing. That’s how you do it in the 21st century.
C.M. MayO: “I can warmly recommend their trade book contract guide, for anyone who might need a light on their path through that particular jungle.”
That’s why I joined. That guide helped me out before I was eligible for the membership.
Anonymouz # 353:
If the current US government is trying to do an organized retreat from its “hyperpower” status, it seems IMHO a very bizarre retreat. Is it really a random fact that Venezuela and Iran alike are oil rich countries? Of course, maybe there isn’t an only motive for attacking Iran (for example, Israeli lobby rules, too).
Michael, thanks for the data point! I expect to see many other companies following suit.
Lunar, here’s hoping — but I doubt it.
Karim, it’s not exactly a last-ditch effort, but it has everything to do with resources. The whole point of the regime-change operation in Venezuela was to get Venezuelan crude flowing to US refineries; my guess is that the point of the Iranian operation is to get a government in Tehran that will provide a fig leaf for the end of sanctions and a flood of Iranian oil and gas into world markets. As the fracking boom winds down, some such gimmick is essential if a sudden supply panic is to be avoided. Of course both those moves simply kick the can further down the road, but that’s all anybody’s doing at this point.
Chuaquin, exactly. The more mass, the more fuel you need to get the craft out of Earth’s gravity well, to bring it into Mars orbit, and then to manage the return journey. The resulting costs would be (quite literally) exorbitant.
Fred, Mike Adams has been wrong about so many things so reliably that I take this as evidence that in fact LLMs will never achieve sentience. Fun fact: my late wife was for several years the editor of an alternative health news bulletin circulated mostly to people in the biofeedback industry. Adams’ dubious predictions were a common theme of amused discussion at our dinner table. Still, by all means believe the sources you prefer; I’d simply encourage you to keep track of how their predictions turn out.
Great Khan, I wish I could say I was surprised. All the more reason for me to figure out some way to insert a “no AI” label in my books. Hmm…
Patricia M, I haven’t read that book, and obviously I need to change that. Thank you!
Other Owen, here we go. Wheee!
Chris, oddly enough, I was thinking about those two words also.
Siliconguy, yep — and that’s another reason I expect massive pruning of bureaucracies as we proceed.
Anonymuz, two excellent points. Yes in both cases; the business in Iran right now is an example of the Chinese proverb, “kill a chicken to scare a monkey” — that is, you beat up on a weak opponent to get stronger ones to back off. As for AI as the Afghanistan withdrawal of the PMC, that strikes me as square on target.
GlassHammer, we’ll see. My take is that we’re on the brink of a massive structural change, but events will show whether I’m right or wrong.
Rajarshi, I hadn’t remembered that Naram-Sin exercised lordship over Meluhha! That would certainly explain some of the extensive cultural exchanges between Mesopotamia and India in very ancient times.
@Fred RE : AI intelligence
“He asserts that the systems are already exhibiting inexplicable levels of intelligence, in that they know things they have never been trained in.”
Recently there was a very interesting project, one that I approve of regarding LLM models. Basically they did the usual training but only on all available texts up to 1913. It is neat to see how it reflects the social norms of the time, but related to the earlier claims about growing intelligence, it was impossible to get it to even point anywhere near any scientific discoveries of the next century. Relativity, quantum mechanics, computers, jet travel, DNA double helix etc it simply couldnt get there. No surprise there. As far as we know, you cannot create intelligence by throwing more data at the prpblem, it is an intrinsic thing that we dont understand. The brain trying to understand the brain, we are at least several millenia into that hole and have a long way to go. It is as Alan watts used to say “Like trying to touch the tip of your finger, with the tip of the same finger”.
Alas, the claim of “1-2 years” until super intelligence is something I have been hearing since the 90s. It is computings “Fusion is 10 years away!” Meme. I will believe it when I see it but too many people out there are desperate for it to be true to have some 1960s scifi come true rather than just look at what is actually happening.
Sorry, this is one of the ugliest websies I have seen in a long time and it very tech heavy but it there is some more context there.
https://devops-geek.net/devops-lab/history-llms-training-language-models-on-pre-1913-texts-a-fascinating-deep-dive-into-temporal-ai/
“The whole point of the regime-change operation in Venezuela was to get Venezuelan crude flowing to US refineries; my guess is that the point of the Iranian operation is to get a government in Tehran that will provide a fig leaf for the end of sanctions and a flood of Iranian oil and gas into world markets.”
That may have been the original point, but that has massively backfired for the Americans. Iran has bombed oil refineries in multiple Arab petrostates and shut down the Strait of Hormuz and all of a sudden most of the world is facing oil shortages and sharply rising oil prices.
Also fertilizer shortages arising from the closure of the Strait of Hormuz, which is a huge concern for industrial agriculture around the world:
https://www.wisfarmer.com/story/news/2026/03/10/u-s-farm-groups-urge-action-as-fertilizer-ships-sit-idle-in-gulf/89073201007/
https://www.foodingredientsfirst.com/news/hormuz-crisis-food-ingredients-supply-chain.html
https://farmpolicynews.illinois.edu/2026/03/attack-on-iran-could-disrupt-global-fertilizer-markets/
#JMG (#361):
All of Hoffer’s books are well worth reading, or so I suppose who have not yet read all of them. He writes in aphorisms, mostly; and by and large, his aphorisms strike me as finely distilled wisdom.
He seems to have had a very challenging life, starting out as a homeless lad of 18 on skid row, then earning a living as a longshoreman for decades, and ending up in a comfortable old age. I view him as something of a very sharp knife, honed to exquisite keeness on the rough and fine whetstones of life.
Rationing of oil has already started happening in many countries in Asia:
https://www.abc.net.au/news/2026-03-10/petrol-prices-surge-asia-middle-east-war-iran-us-israel/106432518
JMG # 361:
No argument here. Real manned long distance space travel is a problematic thing due to radiations (beyond the “easy” sci fi space travel), between another problems.
————
Michael G. # 362:
An interesting comment (link included), thank you. I think some people want to believe in LLM “real” intelligence because they’ve got a god-like Progress Faustian strong bias. To some extent, they behave like people who want to see forms, faces, castles and another imaginative things in clouds (though in a less naïve way). They see what they like to see in LLMs because they need to believe it.
———————————
I’m making an effort for not going toward the off topic comment, but the current Middle East war(s) are influencing me like I guess it happens with you all. It’s near unavoidable. Well, I think there are actually too many “X” in the war(s) equation to grasp an accurate panoramical of the thing. However, I’m going to point that, in addition to the mysterious several “X” I said in the last Open Post, there’s another last “X” to think about it.
I’m thinking about LLM (“AI”) real military use during the current war against Iran. It’s soon to grasp how useful or useless has been this supposed progress in nowadays war, but I’m open to every hypothesis, though I also know it’s difficult to understand well LLM effects on war, due to the evident propaganda from both sides.
Mr. Greer, re. your response to PtKoP.. Perhaps one of those symbols (a circle with a slash through it) depicting a robot as the dreaded NO! .. with something to the effect of ‘For the Discerning Mentat Crowd’ worked around the border. Something small enough to fit unobtrusively on the bookcover.
“Karim, it’s not exactly a last-ditch effort, but it has everything to do with resources. The whole point of the regime-change operation in Venezuela was to get Venezuelan crude flowing to US refineries; my guess is that the point of the Iranian operation is to get a government in Tehran that will provide a fig leaf for the end of sanctions and a flood of Iranian oil and gas into world markets. As the fracking boom winds down, some such gimmick is essential if a sudden supply panic is to be avoided. Of course both those moves simply kick the can further down the road, but that’s all anybody’s doing at this point.”
What the United States achieved with the Iran war will probably be an economic depression or recession.
My hypothesis is that there will be about two or three years of an economic recession, after that economic recession comes the last great effort of this civilization, which can be summarized as “We must make the energy transition, technology will provide.”
Want proof? Look at the people who are still waiting for the lunar bases.
Look on the bright side, solar panels will probably come down in price.
It occurs to me that “no AI” certification could become the next green certification.
The bureaucracy is high in my area, according to the ReCover California consultant I talked to last month, I am in the worst county in the state for dealing with the building department. They do this every day all over the state, but they are done doing it here, their solution is to take any recommended work that requires dealing with the building department off the plate for the homes in the county attempting to utilize the program.
My middle child works locally in a job that did not exist a few years ago, absolutely a bureaucratic job, not only did that position not exist, but then they expanded the scope and hired more people to work for my offspring ! It seems that everyone expects more narrowly defined expertise, the job of course encompasses alot of CYA ( cover your behind), then the general public expecting more… presence ? .. more arent you doing something about xyz. But, also, in my opinion, the biggest reason is alot of other employees are not competent to do what people in their position in the past used to be capable of doing. Instead of insisting on competence over a range of work objectives, the solution was to make a new department. And as soon as my offspring was shown as competent and able, more employees would ask for these parts of their work to be done. It seems to me that we have more and more office workers with alot of slack time and reduced responsibilities ( for some). Like mentioned above at a school the vice principal might have used to be the athletic director also, but now the position has less responsibilities. Of course, in my offsprings case, it was true that all these various other employees were not competent to do anything more than the most basic on their own. A sad state of our public education system.
While my offspring feels an amount of job security from being indispensable, also knows as we all know that hiring and firing doesnt follow competency. I see the increases in these types of positions as not being critical to the organizations mission, as we are discussing on this thread. I would also recommend that anyone who has the room in their lives, to cross train away from lenocracy jobs. This is my offspring that is most vulnerable in our changing world. My worry isnt the particular type of position not existing in 5 or 10 years, it is the physical ( living situation) and psychological toll. I believe that my various offspring, including this one in the PMC position, are innately capable, can do lots of things. But, I also know as I have seen laid off workers in previous downturns ( 2008 for example, before my offsprings time) that people are human and dont pivot as quickly as they should and also can be shocked or otherwise mentally debilitated by changes. For those here that think about this future, I would say to remember being able to change quickly and roll with the personal disaster changes to be something to train for, if this is possible. You have to hold on tight for the ride
Unfortunately, I think you’re right about the necessity of moving, especially with the data center scheduled to come online in 2027. They released an article today regarding the issue, and the water authority (GLWA) is blaming the electricity provider (DTE), and both are going to be put under more stress.
This article does not fill me with competence, nor does the deer in the headlights look of the person responsible.
https://www.candgnews.com/news/glwa-leader-addresses-novis-water-main-issues-with-city-council-10448
@ Robert Mathiesen and @ JMG
Re Eric Hoffer – his early life is even more interesting than that, Robert. 🙂 https://en.wikipedia.org/wiki/Eric_Hoffer
Two details that stood out to me include the loss of his sight (at age seven, after the death of his mother) and the regaining of his sight (at age 15)… and the fact that his father’s death expenses were paid by one of the “friendly societies” that JMG has written about.
Oh, and one other thing… throughout the years he wandered, living as a hobo, and taking casual jobs here and there, he maintained library cards in all of his commonly visited towns. He was an incredibly voracious reader, all his life.
For those who watch video, this links to part 1 of 5 of an interview with Eric Hoffer recorded in September 1967 – https://www.youtube.com/watch?v=kTcv4HyEY3w (parts 2 to 5 will appear in the suggestions box to the right hand side).
A truly original and thoughtful man…
re: AI-washing
The tech industry goes through a 20 year cycle (or scam, depending on how you look at it). They hire a bunch of fresh faced grads who don’t know anything but are willing to work long hours and put up with whatever management says. I suspect it’s the last one they really care about, although performative suffering gets middle management off too.
They beat those people like rented mules for 20 or so years until they get too expensive and experienced and they get too good at working the corporate system (which is another way of saying too expensive). They always find some excuse (and it really doesn’t matter what it is) but they like to shovel them all out the door, all at once. Then they lay low for a few years and then quietly hire – a new crop of fresh faced grads. And try to outsauce as much as they can from the previous cycle. Claude, saar.
And then the cycle starts all over again. Only go into tech if you were that nerdy kid that wanted an arduino or electronics set. Otherwise they’ll chew you up and spit you out. The nerdy kid will still have a tough time of it too, but he was sent here to go through it. And he’ll probably do OK in the end. If you weren’t that kid – avoid, do something else.
Answer to Zemi from March, 7th : we may not live in a simulation but in a distorted timeline, which would explain the Mandela effect and other anomalies, if we consider that reality is constantly edited, just like a movie, by entities from the 4th and the 5th dimension who have the ability to move through time the same way we move through space… We’re trapped in time, being in 3D, but this transdimensional beings aren’t, and it’s then very easy for them to manipulate mankind’s history…
“If the current US government is trying to do an organized retreat from its “hyperpower” status, it seems IMHO a very bizarre retreat. ”
Actually no, it’s a well run retreat. Limited counterattacks keep the opposition off balance and distracted.
I did a google search on “role of counterattacks during a retreat” and got this summary
“Counterattacks during a retreat are essential tactical maneuvers used to break the enemy’s momentum, protect the withdrawing force from becoming a disorganized rout, and regain the initiative. By forcing the pursuing attacker to pause, take cover, or re-organize, these counterattacks allow the retreating army to safely disengage, reorganize, and potentially establish a new, stronger defensive line.”
There was a long list of links afterward and a complete book on the topic. Retreats are difficult. Turning and running (a rout) just gets the army slaughtered.
The United States can blame Israel and the neocons / Zionists for the failure of an Iran war and use it as an excuse to get out of the Middle East. The only problem with this for some commenters on this blog is that Trump has been very pro-Israel in both his terms as president and so Trump might also get some of the blame for the Iran war and become very unpopular among Americans.
@ JMG
“my guess is that the point of the Iranian operation is to get a government in Tehran that will provide a fig leaf for the end of sanctions and a flood of Iranian oil and gas into world markets.”
I don’t see how ensuring the near/mid-term availability of oil lines up with the US government’s behavior here. Over the past year the US has tightened sanctions on both Russia and Iran, has blown up (sometimes lliterally) multiple attempts at negotiations with Iran that could have given sanctions relief in exchange for concessions, and has now entered into a war that would predictably have huge oil supply disruptions that might last for years. Oil supply doesn’t seem like even a secondary consideration in all this.
Greetings, Archdruid!
I’ve been away for quite a while … I’m gonna cut to the chase and say that in everything, large and small, where my thoughts and opinions have differed from yours, yours have proven to be correct and mine have proven to be … well, uninformed, naive, even puerile.
Ah, the 70s. I was a 70s kid. I guess you had to have been there (to the youngins in the room) I was born in 1962, making me about 8 when that notorious decade started, and my life was pretty much 1960s-normal/”normal” and reality set in around 1975. By that time we were being foreclosed out of the last house we’d ever own, we were all skinny due to not enough calories, and it finally got to the point where we were cooking – when we had food – in a fire ring out on the lawn, because we had no electricity.
My father had been stupid enough to bet he could make a living programming computers. No more need be elaborated on how stupid that idea is.
1976 found my mother and we kids, my father having done the standard American thing and run off, and my mother got us Social Security numbers, got us on welfare and food stamps, WIC, etc. We moved down out of the hills and down by the shore where the fishing and foraging were much better. I caught sight of myself in a store window and saw the big belly and pipestem arms and legs and recognized myself in pictures I’d seen in National Geographic. I needed to get more food and quick. I developed decent fishing skills and found food sources that others didn’t want to bother with (fallen coconuts, odd fruits) or things not generally recognized as food (sweet potato leaves) and scrounged for every odd job I could find. By the time I was 14, a good say might be my having weeded a lawn or done some job like that, come home and tell my younger sisters I have $2 or $3, what do we all want for dinner?
It was a time of great nostalgia for the 1930s, with TV shows like The Waltons, and many of the buildings unchanged since the 1920s or 30s. A movie theater had matinee seats for 39s. 49c got you a hot dog at a place downtown, “Your Own Bun”. Bus rides were 10c if you were a kid, 25c for an adult. Prices crept up by the time the 80s came along, but this was the reality. People just struggled along, and if you didn’t have money for the laundromat or the electric had been shut off, again, you got out the washboard and did the laundry.
So when the economy started back onto the path of normalcy in 2008 and I lost everything, I was able to tell myself, “Well, I’ve done this before, and I’m bigger and stronger and better-fed this time, and know a few more tricks”. And we’re headed for another big dip so the rich can rake off anything they can, and it’s gonna be a real trauma to those who did not survive the 70s because it’s a very different mode of life. We managed to keep one phone going between 3-4 houses, will the kids get used to pooling their money and keeping one phone alive between a handful of them?
@ Atmospheric River #370
Those are “ouchy” data points!
When you say – “the biggest reason is a lot of other employees are not competent to do what people in their position in the past used to be capable of doing. Instead of insisting on competence over a range of work objectives..”
I would mostly agree, but with a subtle change of emphasis. Because the competence that you used to find so much more frequently, within the employees embedded in state or corporate hierarchies, arose from experience, and to gain that experience they had to be trusted to take action, as well as record and take note.
In my experience, 20 years ago, I could easily find *someone* within the tax office, or the customs office, or within the customer service department of a corporation, who would take ownership of whatever problem or difficulty you were bringing to their attention. They might not be able to solve it, but they would take it upon themselves to go to whatever internal people they had to, or do research into applicable policy, until they could. I could keep that person’s name, direct phone number, and direct email address, and remain in communication with THAT person as my one stop shop to dealing with that organisation.
Somewhere, somehow, over the past 20 years all of those people were de-entrusted with any power to take ownership of problems the public bring to them. The public-facing numbers and emails are not individually staffed, but answered by bots and/or phone bank staffers. They will (mostly) listen politely without any power to DO anything, and should you get a glimmering of a solution in speaking to an actual person, you will never be able to speak to THAT person again, since you will be back to square one the next time you call, and have to explain the whole thing yet again to some new person.
Competence comes with the exercise of power, through which one gains experience. At the same time that bureaucracies were multiplying their employee numbers, they were diminishing their powers to act, and thereby gain competence.
“I don’t see how ensuring the near/mid-term availability of oil lines up with the US government’s behavior here. Over the past year the US has tightened sanctions on both Russia and Iran, has blown up (sometimes lliterally) multiple attempts at negotiations with Iran that could have given sanctions relief in exchange for concessions, and has now entered into a war that would predictably have huge oil supply disruptions that might last for years. Oil supply doesn’t seem like even a secondary consideration in all this.”
The big elephant in the room here is the Epstein Files. Trump started a war with Iran to distract everybody from the horrifying revelations about Trump in the Epstein Files, which Trump was trying to prevent from releasing but was forced to release by MAGA congresspeople like Thomas Massie, Marjorie Taylor Greene, and Nancy Mace.
“Outside of a narrow range of uses, [Most Humans] perform very poorly. ”
Do you remember the function of the Gom Jabbar in ‘Dune’? As my AI reminds me: “The name “Gom Jabbar” is derived from Arabic roots—Gôm (enemy) and Jabbâr (tyrannical)—meaning “a high-handed enemy.” It symbolizes the existential pressure that drives evolution and growth. ”
The sad truth is that few humans can pass the Test of being in any more alive than an AI or a Non Player Character. I think we need to acknowledge that and, for ourselves, prove we can escape the box.
“After the tricycle it comes always the bicycle.”
As a software engineer for a large tractor company, i feel like my role is partly bureaucratic, it relies on a drive to add software complexity which assumes sw is needed on tractors. There is quite a bit of savings with that sw since it allows users to save on seed and spray since they can know where they applied those on the field to prevent double or no application but is that worth the cost of all the complexity costs added? is the question. Always interesting to try and guess where on the totem poll of ROI you are and how close your field is to negative ROI assuming complexity costs will go up.