I'm worried about AI psychosis. Specifically, I'm worried about the psychosis that makes "capital allocators" spend *$1.4T* on the money-losingest technology in human history, in pursuit of a bizarre fantasy that if we teach the word-guessing program enough words, it will take all the jobs.

--

If you'd like an essay-formatted version of this thread to read or share, here's a link to it on pluralistic.net, my surveillance-free, ad-free, tracker-free blog:

https://pluralistic.net/2026/04/13/always-great/#our-nhs

1/

That's some *next-level* underpants-gnomery:

https://pluralistic.net/2026/03/12/normal-technology/#bubble-exceptionalism

The thing that worries me about billionaires' AI psychosis isn't concern for their financial solvency.

2/

Pluralistic: Three more AI psychoses (12 Mar 2026) – Pluralistic: Daily links from Cory Doctorow

No, what I worry about is what happens when the seven companies that comprise a third of the S&P 500 stop trading the same $100b IOU around while pretending it's in all of their bank accounts at once and *implode*, vaporizing a third of the US stock market.

My concern about a massive collapse in the capital markets isn't that workers will suffer *directly*.

3/

Despite all the *Wonderful Life* rhetoric about your money being in Joe's house and the Kennedy house and Mrs Macklin's house, the reality is the median US worker has *$955* saved for retirement. You could nuke the whole financial system and not take a dime out of most workers' pockets:

https://finance.yahoo.com/news/955-saved-for-retirement-millions-are-in-that-boat-150003868.html

4/

$955 saved for retirement? Millions are in that boat.

The typical American worker has less than $1,000 saved for retirement, according to a new report from the National Institute on Retirement Security (NIRS).

Yahoo Finance

No, the thing that has me *terrified* about AI is that when it craters and takes the economy with it, that we will respond the same way we have during every financial crisis of the 21st century: with austerity, and austerity breeds *fascism*.

There's a direct line from every K-shaped recovery to every strong-man who's currently sending masked gunmen into the streets.

5/

The Hungarian dictator Viktor Orban rose to power after people who'd been suckered into denominating their mortgages in Swiss francs lost their houses when the currency markets moved suddenly, because the swindlers who'd sold them those mortgages took the position that wanting to live somewhere automatically made you an expert in forex risk, so caveat fuckin' emptor, baby.

6/

Back in America, Obama decided to bail out the banks and not the people. His treasury secretary Tim Geithner told him the banks were headed for a catastrophic crash and could only be saved if he "foamed the runways" with everyday Americans' mortgages. Millions of Americans lost their homes to foreclosure as banks, flush with public cash, threw them out of their homes and then flipped them to investment banks who became the country's worst slumlords:

https://pluralistic.net/2022/02/08/wall-street-landlords/#the-new-slumlords

7/

Pluralistic: 08 Feb 2022 – Pluralistic: Daily links from Cory Doctorow

Americans were understandably not entirely happy with this outcome. So when Hillary Clinton replied to Donald Trump's "Make America Great Again" with "America is already great," her message was, "Vote for me if you think everything is great; vote for Trump if you think everything is *fucked*":

https://www.politico.com/blogs/2016-dem-primary-live-updates-and-results/2016/03/clinton-america-is-already-great-220078

"Austerity begets fascism" is one of those things that makes a lot of intuitive sense, but it turns out that there's a good empirical basis for believing it.

8/

Clinton: America is already great

Clinton also congratulated Sanders on his win and thanked her supporters and allies.

POLITICO

In "Public Service Decline and Support for the Populist Right" four economists from the LSE and Bocconi provide an excellent look at the linkage between austerity and support for fascists:

https://catherinedevries.eu/NHS.pdf

Here's how they break it down. Political scientists have assembled a large, reproducible body of evidence to show that "public service provision is crucial to people’s perceptions of their quality of life and living standards."

9/

Good public services are the basis for "the social contract between rulers and the ruled" - pay your taxes and obey the laws, and in return, you will be well served.

When public services go wrong, people don't always know who to blame, but they *definitely* notice that something is going wrong, so when public services fail, people stop trusting the state, and that social contract starts to fray.

10/

They start to suspect that elites are lining their pockets rather than managing the system, and they "withdraw their support" for the system.

Fascists thrive in these conditions. Fascists come to power by mobilizing grievances. By choosing a scapegoat, fascists can create support from people who are justifiably furious that the services they rely on have collapsed.

11/

So when you can't get shelter, or health care, or elder care, or child care, or an education for your kids, you become a mark for a fascist grifter with a story about "undeserving migrants" who've taken the benefits that should rightly accrue to "deserving natives."

12/

(This is grimly hilarious, given that the wizened, decrepit rich world is critically dependent on migrants as a source of healthy, working-age workers who pay massive amounts into the system while barely making use of it, many of whom plan on retiring to their home countries when they do reach the age where they're likely to extract a net loss to the benefits system.)

13/

Enter the NHS, a beloved institution that is hailed as the pride of the nation by both the political left and the right. The majority of Britons use the NHS, with only 12-14% of the population "going private," so when the NHS declines, *everybody* notices (what's more, even people with private care use the NHS for many of their needs).

Britons love the NHS and they want the government to spend more on it.

14/

There's "a broad public consensus that the government is not going far enough when it comes to funding." That's because generations of cuts to the NHS have left it substantially hollowed out, with major parts of the service handed over to for-profit entities who overcharge and underserve.

15/

The most tangible and immediate evidence of this slow-motion collapse comes when your local general practitioner ("family doctor" or "primary care physician" in Americanese) shuts down. The UK has lost 1,700 GP practices since 2013.

Reasoning that a GP closure would make people angry at the system, the economists behind the paper wanted to see what happened to people's political beliefs when their GP's office shut.

16/

They relied on the GP Patient Survey, a longitudinal study run by NHS England and Ipsos Mori. The survey asks a statistically significant random sample of patients from every GP practice in the NHS and then weights the results "to reflect the demographic characteristics of the local population according to UK Census estimates." It's good data.

17/

The researchers cross-referenced this with various high-quality instruments that measured the political views of Britons, like the U Essex Understanding Society Panel, drawing on 13 years' worth of surveys from 2009-2022, gaining access to a protected version of the dataset with fine-grained geographic information about survey respondents, which allowed them to link responses to the "catchment areas" for specific GPs' office.

18/

They combined this data with the British Election Study panel, which has surveyed voters 29 times since 2014.

Most of the paper describes the careful work the researchers did to analyze, cross-reference and validate this data, but what interested me was the conclusion.

19/

People who see a severe degradation in the quality of the services they rely on switch their political affiliation to one of Britain's fascist parties - UKIP, the Brexit Party, or Reform - parties that have called for ethnic cleansing in Britain.

This is what has me scared. We can see the looming economic crises in our near future. If it's not the AI crash that triggers the next wave of austerity, it'll be the oil crisis created by Trump's bungling in the Strait of Epstein.

20/

And of course, we could always get a twofer, because the Gulf States that *were* pouring hundreds of billions into AI data-centers now need every cent to rebuild the LNG shipping terminals and oil refineries that Iran blew up after Trump, Hegseth and Netanyahu started murdering all the schoolgirls they could target. Once they nope out of the AI bubble, that could trigger the collapse.

21/

@pluralistic
was the measure, the target?

@pluralistic

And Ripe for the privatization carpetbagging by the likes of Pete Theil and Palintir. ... I mean Private Capitalization..
because capitalists gonna CAPOTOLIZE

@pluralistic bears repeating (and is quite literally true in the uk case, where reform is full of ex tories and bankers) the the people who caused the crash and implemented austerity are THE SAME PEOPLE as those pushing fascism

@pluralistic

Doesn't *START* to fray, it continues to fray or frays more.

(having difficulty with fraysing.)

@pluralistic

Are you refering to forex the condom or forex the market...
... thinking...
Maybe there isn't much of a difference both are used for screwing.

@pluralistic Linked article states $955 is the median value (so, 50% of workers have less), not the 95th percentile value.

@pluralistic

I have been describing AI as a business cult. CEOs signing up to spend mountains of money, jettisoning any analytical discipline. I chalk it up to psychosis born of unchecked monopoly. They're all at the Davos circle jerk getting played by the most sociopathic grifters among them.

What could go wrong?

@jawarajabbi @pluralistic
Leveraging FOMO has always been the key skill of grifters. Never has that been more evident than with the rush to “AI” everything.

@jawarajabbi @pluralistic

It's difficult for CEOs to escape AI when their board or investors are asking how they're going to protect the company from a potentially instant devaluation of their codebase. If the AI bet turns out to work, then a 10 year codebase could lose 50% of its value because software can now be built in half the time using AI agents. Not to mention the commodification of software features that would occur when the barrier of entry to building software is lowered.

@jawarajabbi @pluralistic

It doesn't help that investors are now being pitched POCs that were built entirely using AI agents, which becomes an incentive to put pressure on existing investments to increase productivity using AI. Of course what's missing from that picture is what happens after the POC becomes a product, and security and compliance enter the picture.

@jawarajabbi @pluralistic

So we have developers implementing AI because they're being pressured to optimize, CEOs implementing AI because they're pressured to protect their company and investors buying into AI because they fear the devaluation of their investments.

Some data seems to indicate that companies believe that AI adoption is in its nascent stage, even when their own bets have only paid off marginally, so it looks like this is going to continue for the foreseeable future.

@davidsonsr @jawarajabbi @pluralistic The problem is similar to that of the 00s or the 10s: slapping cloud tags on a proposition to attract investment, followed by slapping blockchain on everything, and now it's GenAI instead. The investors involved cannot be relied upon to take a sufficiently well-researched and scientific approach. They rely on mediated experience to reach judgements as we all do. But it is musical chairs now, given the level of mistruth and misrepresentation now happening.

@bms48 @jawarajabbi @pluralistic

I'd say that this is different because it's not just a marketing gimmick, it's a variable that promises an unknown amount of cost optimization.

For people who have spent their lives amassing a fortune, the prospect of losing a significant portion of it (or all of it) becomes an incentive to spend an insignificant amount of that fortune on AI in order to hedge the bet.

@bms48 @jawarajabbi @pluralistic

One thing to take into consideration is that AI is already replacing human tasks since companies have been embedding AI into their organizations for a long time now, so investors extrapolate on these cost savings and buy into the bet that AI will eventually replace significant amounts of organizational work.

@bms48 @jawarajabbi @pluralistic

I don't really know if this is likely to stop any time soon, but it looks like a lot of people are engaging in discussing the environmental effects of AI which seems to be having some political impact.

@davidsonsr @jawarajabbi @pluralistic The pressure is upon me, choosing self employment after 7 year sabbatical, to adopt GenAI in some form or other. Most of the architectural work I am doing could not be directly assisted by GenAI, and "vibe coding" just does not enter into it. Limited LLM use for new code using specific models. Perhaps LLM assisted code security audit. But outsourcing it all to Claude Code? That is really stupid; my strategy is common with other "chess players". Be a centaur!
@bms48 @davidsonsr @jawarajabbi @pluralistic

Modern "investors" seem to mostly favor quick returns. This puts them ever on the hunt for hockey sticks rather than long term steady performance. The "flipper" phenomena seems to have infected vast swaths of the investment segments of the economy. "Research", now is mostly in finding quick scores and almost actively ignores signs that an "investment" target eschews sounder principles.
@davidsonsr @jawarajabbi @pluralistic There is anecdata for the beginnnings of a pushback from professional risk control folk, credit reference agencies, reinsurers.

@bms48 @jawarajabbi @pluralistic

My suspicion is that AI companies will attempt to lobby for reduced compliance requirements with software built using AI, but I'd expect some pushback on that from regulators.

@davidsonsr @jawarajabbi @pluralistic There has already been, pre-dating and in parallel with the self-delusion of "vibe coding" (unless you consider throwaway prototypes of web CRUD apps the end goal vs actual engineering), a push back towards proper security compliance and audit, and assignment of liability in the software industry: https://cacm.acm.org/practice/the-software-industry-is-still-the-problem/ Reinsurers, credit agencies, risk control professionals, are starting to push back against GenAI proponents. @bsdphk
The Software Industry Is Still the Problem

Communications of the ACM

@bms48 @davidsonsr @jawarajabbi @pluralistic

Not to mention EU's revision of the Product Liability Directive, a revision they undertook only and specifically to apply "no-fault" product liability to software in the consumer market.

Coming to your EU-country later this year...

@bsdphk @davidsonsr @jawarajabbi @pluralistic
I snagged a high-level guide to the updated EU directive from... an insurance firm. Told ya! "“medically recognised damage to psychological health” is now also expressly included here."
I think the EU have identified Meta in their threat model and made it part of their legislative framework.
Right On Commander.
https://www.hdi.global/globalassets/_local/international/downloads/group_product-liability-directive/HDI_Factsheet_Liability_NewProductDirective_INT_en.pdf
@jawarajabbi @pluralistic the "Davos circle jerk" is at the same time a nightmare-inducing and perfectly accurate image
@pluralistic Absolutely this. I'm sure there's a straight line to be drawn from the 2008 financial crisis (and particularly the government's response of austerity) to the 2016 Brexit vote; through to today's horrifying position of Reform potentially coming into power in a couple of years. 18 years of declining public services, flatlining wages & visible corruption of parliament (most apparent under Johnson) has left the people bitter & angry.

@gsquirrel @pluralistic

It's so obvious it's under our noses.

Mandelson and Epstein circa 2008

George Osborn, Chancellor from 2010 onwards - photographed partying on Epstein's yacht.

Both main parties captured by this group of paedo billionaires, all agreeing austerity is great.

@dar @gsquirrel @pluralistic
"People as things, that’s where it starts."
(Carpe Jugulum)
@dar @gsquirrel @pluralistic On 10 May 2010, Mandelson revealed to Epstein the existence of a secret underground tunnel between 10 Downing Street and the Ministry of Defence. https://www.printernational.co.uk/timmann2/history.htm#epstein
Max Spiers's Chronicle

Max Spiers's Chronicle

@gsquirrel @pluralistic Something similar in Italy.
That, and turning words like "socialism" into political slurs, for some fucking reason.

@pluralistic

This technology needs rules and regulation before being set loose on the public.

But big tech is only seeing the $$$

What could possibly go wrong.

LLM can be useful but only in the right hands with the right skills, and is exactly skills we will lose is this is used in the present scheme of things.

The WWW has democratized access to information (but also shows the same tabulation as it is also an accelerator for nonsense).

LLM now will take that away, not what we need!

@xs4me2 @pluralistic There is a reason why all the Tech bros went absolutely bonkers after Trump was elected - they knew perfectly well that he and his gouvernment won't regulate anything if they lick his boots nice enough.
@[email protected]

Wir leben in einer Welt, in der es mit Wikipedia das Großartigste gibt, was je von Menschen geschaffen wurde, eine Kollektion von großen Teilen des Wissens der Welt. Und gemacht worden ist es von Menschen, die dafür nicht bezahlt worden sind und für praktisch überhaupt kein Geld.

Auf der anderen Seite haben wir Chatbots die den größten Unsinn überhaupt verzapfen und das für das meiste Geld, das je ausgegeben worden ist.

Und alle Menschen haben die freie Wahl, entweder bei Wikipedia mitzumachen oder Geld in Chatbots zu stecken.

Und die Menschen machen nicht bei Wikipedia mit sondern stecken jede Menge Geld in Chatbots. Alle 8,3 Milliarden. Menschen oder Dollar oder was auch immer.

@pluralistic Fascism – what Hannah Arendt called 'organized loneliness' – can only take root when people stop believing that their society will reward their lawfulness with an orderly and humane existence.

Thanks for that Hannah Arendt quote. TIL.

#Fascism #Collapse #Austerity

@pluralistic I hear what you're saying and it's broadly right, but I want to push back a little bit. I think you're looking at a symptom and not the disease. If AI undeniably made the life of the masses better, we wouldn't object to it. But that's not what's happening, so what are we objecting to? It's extraction: when the wealthy use their power to acquire wealth instead of generating it.[1]

1. https://fosstodon.org/@ovid/116334866923361500
2. https://curtispoe.org/projects/extraction/

@pluralistic
I've built The Extraction Index to help us track this worldwide. ([2] in previous post). This behavior has been documented to occur for thousands of years across many civilizations and economic systems. Over the millennia, extractive behavior has become less severe individually (e.g., we outlaw slavery), but it becomes far more widespread.

If we keep treating the symptoms instead of the disease, we're playing whack-a-mole: the wealthy will always find a way to abuse power.

@pluralistic 🧵 Reading your part about the NHS, makes me (in France) realise: I'm not worried that any billionaires could have an "AI" psychosis (then they could get therapy and medicaments).
I'm deeply worried about mentally ill people in a *real* psychosis who don't get therapy because not enough specialist doctors, support services, places. Whose families are already being lulled into a false sense of security by the idea that AI-powered counselling and therapy will surely be available soon.

@pluralistic 🧵 We are all aware of the deadly consequences that already happened with AI chatbots.

And that brings us back to ‘good old' eugenics. I wouldn’t call that "AI psychosis", but rather, quite plainly: a cult of death, a fascist ideology. There are no pills to counter that.

@NatureMC @pluralistic I could be misreading, but I think the idea is not that (only) a handful of powerful people are affected by "AI psychosis." This is also what is emerging as a symptom in regular people who now rely on AI as a babysitter, financial advisor, intern, etc. and ultimately believe the AI over the people around them. There are increasing cases of AI reinforcing people's delusions of any nature, more rapidly and consistently than any human could, until they essentially break from reality. Most visibly, AI has been fueling people thinking they are misunderstood geniuses and making it very hard to being them back to a state where they can be reasoned with.

The added danger of it happening to extremely powerful people is a consequence of their power, and thus affects us all via the ways they shape the world we live in. Delusion-driven acts on this scale are not new, but this is a new source of them.