AI is driving some people mad

Some individuals seem to be particularly susceptible to AI psychosis, which shouldn’t be surprising. Humans have long been prone to fall for cults and authoritarian leaders, and what is ChatGPT but a machine trained to emulate authoritarian pronouncements with an all-knowing attitude, marketed by some of the worst people on the planet? So a few people have fallen down the rabbit hole.

Until recently, the tech industry and its financial backers have had little to say about the phenomenon. But last week, one of their own — venture capitalist Geoff Lewis, a managing partner at the multi-billion dollar firm Bedrock who is heavily invested in machine learning ventures including OpenAI — raised eyebrows with a series of posts that prompted concerns about his own mental health.

In the posts, he claimed that he’d somehow used ChatGPT to uncover a shadowy “non-government agency” that he said had “negatively impacted over 7,000 lives” and “extinguished” 12 more.

Yeah, don’t believe anything ChatGPT tells you. It’s a psychiatric phenomenon that has been around for a good long while — AI is just poised to exploit it.

“There’s recently been an influx of case reports describing people exhibiting signs of psychosis having their episodes and beliefs amplified by an LLM,” wrote Cyril Zakka, a medical doctor and former Stanford researcher who now works at the prominent AI startup Hugging Face.

“While I’m not a psychiatrist by training,” he continued, “I think it mirrors an interesting syndrome known as ‘folie à deux’ or ‘madness of two’ that falls under delusional disorders in the DSM-5 (although not an official classification.)”

“While there are many variations, it essentially boils down to a primary person forming a delusional belief during a psychotic episode and imposing it on another secondary person who starts believing them as well,” Zakka posited. “From a psychiatric perspective, I think LLMs could definitely fall under the umbrella of being ‘the induced non-dominant person,’ reflecting the beliefs back at the inducer. These beliefs often subside in the non-dominant individual when separated from the primary one.”

Unfortunately, this article quotes Eliezer Yudkowsky extensively, who is not an AI, but has had the same kind of deleterious effect on people as any chatbot. If we killed ChatGPT, we’d still have the same problem, only with self-declared human prophets who tell people nonsense, yet are believed.

I am either amused or horrified that the poor deluded victim of AI psychosis is citing the SCP Foundation.

As observers quickly pointed out, the ChatGPT screenshots Lewis posted to back up his claims seemed to be clearly inspired by a fanfiction community called the SCP Foundation, in which participants write horror stories about surreal monsters styled as jargon-filled reports by a research group studying paranormal phenomena.

Jeremy Howard, Stanford digital fellow and former professor at the University of Queensland, broke down the sequence that led Lewis into an SCP-themed feedback loop.

“When there’s lots of training data with a particular style, using a similar style in your prompt will trigger the LLM to respond in that style,” Howard wrote. “The SCP wiki is really big — about 30x bigger than the whole Harry Potter series, at >30 million words! Geoff happened across certain words and phrases that triggered ChatGPT to produce tokens from this part of the training [data].”

“Geoff happened across certain words and phrases that triggered ChatGPT to produce tokens from this part of the training distribution,” he wrote. “And the tokens it produced triggered Geoff in turn.”

The SCP has a clear statement about their intent and purpose.

The SCP Wiki is a collaborative speculative fiction website about the SCP Foundation, a secretive organization that contains anomalous or supernatural items and entities away from the eyes of the public — or so it may appear.

A fictional conspiracy website of 30 million words is part of the training data for AIs that have no context, no ability to distinguish fact from fiction, which then credulously fills gullible human minds with bullshit. That’s the real danger of AI.

I wonder if the AIs have also been soaking in Charles Stross’s novels. I wouldn’t be surprised. It’s not a purely AI thing.

Actually, throughout history people’s minds have been poisoned by the Book of Revelation. Ban the Bible!

Have you ever heard of BBM?

That’s Brown Brothers Media, and you’ve probably read some of their articles. They produce Space Daily, which I’ve encountered a few times, and a hodge-podge of other online ‘journals’, like Scandinavian Standard, Silicon Canals, Sentient Blogs, The Law Dictionary, Science Blog, Mars Daily, Solar Daily, and Nuclear Power Daily, as well as a lot of squishy pop psychology sites, and even one titled Bible Scriptures. They are indiscriminate in the topics they claim to cover, and churn out a tremendous volume of articles that they feed into the Google mill.

How do they do that? They’re based in South East Asia, and they started by hiring low-cost regional writers who could churn out articles with little research, but then they ended up firing most of them and going even cheaper.

“More and more writers became robot writers,” the former contractor said, “until it came to a point where everyone’s just a robot writer, essentially.”

At first, she told us, the robot writers were scrambling to “create our own prompts” to feed into ChatGPT, or were provided with prompts that the writer understood to have been created by the Brown brothers. By 2024, BBM was so invested in the tech that it rolled out its own internal AI tool. It was built into the company’s WordPress backend, the former writer told us, where it could generate entire articles straight into the content management system — with a series of options that let staff quickly specify what type of piece the AI should spit out.

“They started developing this AI tool that generates articles in a couple of minutes,” the former writer explained. “You just select the topic it belongs under (like zodiac, love and relationship advice, etc) and then you’d select the tone (friendly, informative, etc) and it generates an article for you.”

The Brown brothers, particularly Justin and Lachlan, have personally bylined prodigious numbers of articles for their publishing empire, where they typically strike a tone of chin-stroking wisdom about topics ranging from life and spirituality to psychology and society. But according to another former contractor and internal documentation obtained by Futurism, their articles are AI-generated as well, and the company’s AI tool includes a dropdown menu that lets users select the voice of the author in which to generate a piece — including the Brown brothers themselves. (AI detectors corroborate the claim that prior to BBM’s embrace of AI, articles attributed to the brothers were human-written, but are now entirely AI-generated.)

They’re not done. They’re branching out into technology companies that produce more AI dreck.

The Brown brothers also appear to be involved with several AI companies that have marketed AI “scaling” techniques as a service, using language suggesting that they may produce similar content for other media ventures.

Two of these startups, Mosquito Fleet and Tippa Fleet, have United Arab Emirates business addresses and claim on nearly–identical sites to “specialize in using artificial intelligence to improve and scale content production for our partner websites.”

Another company linked to the Brown brothers, Cognitive AI Systems Limited, is registered in the British Virgin Islands. A version of its site archived in February 2025 marketed “state-of-the-art AI technologies to scale content responsibly,” emphasizing that potential partners could “tailor AI capabilities to fit your content production needs, whether you publish daily or by the minute.” By earlier this year, however, that language was gone, with the site’s homepage instead declaring that “Cognitive AI Systems Limited owns a portfolio of digital publications operated by Brown Brothers Media.”

Look forward to even more slop on the web. They are pumping out so much garbage, that they aren’t even bothering to proofread anymore. Why should they? It’s not as if quality matters in their content.

Sometimes it appears that BBM has been so sloppy that it’s affixed the wrong name to a piece generated for a different author. Though Lachlan has written extensively about being married to a woman and having a young daughter, for example, the BBM cofounder recently bylined a May Expert Editor article — titled “10 Signs a Man Is Trustworthy and Has Good Intentions” — in which he mused that “in my personal experience, I knew my now-husband was a keeper when he bravely voiced his insecurities and fears about his career early into our relationship.”

At times, it’s difficult to believe that any human at BBM is taking a single look at drafts before they’re published, or even afterward. Another BBM property, Eluxe Magazine, seems to have accidentally published a pair of articles attributed to Lachlan this summer which consist of nothing more than a few lines of garbled computer code — where they remain pinned to the top of the publication’s home page at the time of this story’s publication.

I guess you can’t trust anything you read on the internet. At the very least you should avoid anything on this list of every known BBM publication.

The intellectual pick-pocket

Imagine I were invited to join a lab, and was given access to their data. I frequently talk with lab members, and I help them with writing up their results. Then, suddenly, I publish a grand paper summarizing their work, claiming that any novel insights were entirely my own, and didn’t give any credit to that generous lab.

Would you accept my claim to full credit?

That’s precisely what Anthropic has done!

When Anthropic, the artificial intelligence giant, unveiled findings from its new biology lab last week, its scientists claimed to have used A.I. agents to discover new enzymes with promise for biotechnology.

Some scientists were quick to cast doubt on the achievement. Then Mario Rodríguez Mestre, a computational biologist, said this weekend that he and his colleagues had been studying the enzymes and their associated molecules — which Anthropic calls ARTs — for four years.

Dr. Mestre and his colleagues have yet to publish their findings. But for the past three years, they have regularly used Anthropic’s A.I. models as they have written code, drafted manuscripts and performed other tasks.

In doing so, Dr. Mestre said he and colleagues shared key findings about the enzymes with Anthropic.

I don’t understand how they can claim that they didn’t suck evidence and arguments from a human science lab — that’s their whole business model! What’s not good for their business, though, is stealing credit, which has a natural consequence.

Dr. Mestre had become increasingly enthusiastic about using A.I. for his science. Just last month, he began using Claude Science, a new Anthropic A.I. tool tailored to researchers.

But after Anthropic’s announcement last week, he is winding down his projects with Claude and shifting to other models.

“I’m shutting down everything,” he said.

Why would anyone type their hard-earned evidence directly into a machine that will blithely steal it?

Biology don’t care

Stumbling across an online conversation between AI advocates is so weird, because they are using the language and concepts of evolutionary biology, but in such a crude, mangled, ignorant way that ultimately, they’ve stripped away the biology to replace it with a religious eschatology. There is a whole fantastic mythology they’ve developed in which they are battling for a paradisial end goal in which they become gods, fused with their technology of Artificial Super Intelligence (ASI) which doesn’t actually exist, but hey, we can measure our progress by looking at NVIDIA’s profit growth.

All they really care about is capitalism’s march into the future, and they’ll ignore all the little details. Build more data centers! If we don’t, we’ll stop all our evolutionary progress. I just want to yell at them that there is no such thing as evolutionary “progress,” and that you can’t change it, you can only vary environmental conditions such that you get a different outcome, and it won’t be the one you expect. They have a belief that they can control how they “evolve,” while not really understanding how evolution works.

So right now they are banging on about human extinction, and how AI can be either a boon or a bane to humanity, and how they need power to shape its direction Of course, that’s what it’s really about: power. They’ve got no idea what they’re doing except that getting legal and financial power over their technology is necessary to keep them rich and happy. It’s all a massive misdirection to whimper about the threat of AI while clutching at the power they want, which is mainly a childish illusion taken from science fiction (see Elon Musk for the premiere example of that.)

Dr. Émile P. Torres is one of the people opposing the nonsense, and I mostly agree with his take on it, but sometimes I feel like he’s so deep into it that he is losing track of how silly these people are. He’s expecting people to recognize how profoundly stupid the debate AI proponents are by attempting to explain their terminology, but then it ends up being an effort to define their terms.

With all this talk about human extinction right now, I am BEGGING people to understand the difference between TERMINAL and FINAL extinction. TESCREALists *only* care about final extinction, but nearly the entire public *misinterprets* them as warning about terminal extinction.

Oy. I always have to look up what stands for — it’s not a very intuitive acronym.

TESCREAL is a neologism proposed by computer scientist Timnit Gebru and philosopher Émile P. Torres. An acronym, it stands for Transhumanism, Extropianism, Singularitarianism, (modern) Cosmism, Rationalism (denoting the ideology of the rationalist community), Effective Altruism, and Longtermism. Gebru and Torres argue that these ideologies should be treated as an “interconnected and overlapping” group with shared origins. They claim these constitute a movement that allows its proponents to use the threat of human extinction to justify expensive or detrimental projects and consider it pervasive in social and academic circles in Silicon Valley centered on artificial intelligence. As such, the acronym is sometimes used to criticize a perceived belief system associated with Big Tech.

It’s a garbage pail word used to hold a lot of garbage, in other words, making it not particularly memorable. It’s also always disappointing because none of the letters stand for “Eugenics,” which is actually what it means, with all the argle-bargle nonsense philosophy assembled to hide that reality.

Torres goes on to explain the distinction between terminal and final extinction.

Terminal extinction is the disappearance of our species, full stop. Final extinction is the disappearance of our species such that we don’t leave behind any (worthy) posthuman successors. Since TESCREALists want a cosmic utopia/posthuman paradise through “aligned” ASI, what
they really mean when they say, “We must avoid human extinction” is: “We must avoid final human extinction.” Once posthumanity arrives, most don’t give a SHIT whether OUR SPECIES persists. Indeed, as I have documented in tremendous detail, many WANT terminal extinction to happen
once posthumanity arrives via “aligned” ASI. If what you care about is our species’ survival — i.e., avoiding terminal extinction — then these people are literal *pro-extinctionists*. DO NOT BE FOOLED by their rhetoric; by the sneaky language games they play!! They are NOT on
Team Human. They are on Team Posthuman. They care about our species surviving long enough to create “aligned” ASI. Once that happens, our species SHOULD die out. These people are not your friends–they are your enemies, insofar as you value humanity. All the quotes can be found
here. Read it and be very terrified that TESCREALists are running the AI companies and monopolizing the conversation. This is so deeply fucked up.

This handy diagram helps explain what he means. It’s a cartoon sketch of a cartoon concept held by a bunch of clowns.

What bugs me about the diagram is that it treats extinction and the branch point as two moments, sharp and discrete, in the history of the species, and ignores the fact that each line is actually a cloud of human variation bumbling its way into the future, and that what happens in extinction is actually a slow sad decline in the face of multiple environmental threats, and the branch point is just the gradual separation of the cloud. With some members of each cloud continuing to interbreed. There would be lots of exchange of genetic material.

In fact, in this longish essay on all these philosophies, the TESCREALists scarcely acknowledge the existence of genes, and seem to think their goal is to shed their genetics and abandon biology altogether.

Pro-extinctionist views go back to the very origins of the TESCREAL movement. In a 1989 article, published around the time Extropianism was taking shape, roboticist Hans Moravec describes himself as “an author who cheerfully concludes that the human race is in its last century, and goes on to suggest how to help the process along.” He claims that we’re developing “intelligent” machines that will soon “be able to manage their own design and construction, freeing them from the last vestiges of their biological scaffolding, the society of flesh and blood humans that gave them their birth.” He then proclaims that “this is the end,” as “our genes, engaged for four billion theirs in a relentless, spiraling arms race with one another, have finally outsmarted themselves”—by enabling the rise of world-dominating digital posthumans. But we shouldn’t fear this transition, he insists, because “post-biological life” in the form of advanced AIS, which he describes as “the children of our minds,” will prosper and flourish.

I hate to tell Moravec the troubling reality that there, beneath the shadow of his crystalline silicon post-human digital paradise, people will go on rutting in the earthy foundation, and that apparently his reality is going to rely on that biological foundation to hold it all up. The genes will go on, since they aren’t going to depend on venture capitalists and vast amounts of energy, water, and rare minerals to keep on going.

To me, it sounds like the AI weirdos are clinging to an outmoded model of orthogenesis, where evolution is linear and progressive, and wouldn’t even recognize the terms anagenesis (the gradual accumulation of variation in a single lineage) and cladogenesis (the splitting of a lineage into two). They’re too busy inventing nonsensical terminologies that aren’t contingent on understanding the reality of the process.

It’s all very silly, except for the fact that these bozos are trying to take over our government and economy.

I may be wrong. I didn’t wait for the STRONG AND SMART (High IQ!) person to weigh in on the matter.

An interesting, but arbitrary, engineering exercise

The World Humanoid Robot Games took place in Beijing a few days ago. They’re fascinating and baffling to watch.

That “Humanoid” in the title is clearly a significant constraint that artificially forces the human creators into making some very bad design choices. But why? No practical devices will be built with that limitation: in the races, for instance, if you’re going to mimic biology, you ought to go for four or six legs, and why bother to imitate an animal, just put wheels on the goddamn thing. What are the advantages of requiring your running robot to clumsily use a bipedal form? I suppose it’s an exercise in building limbs to a difficult and ultimately stupid specification, but why?

It is kind of menacing and intimidating to see a field of soldier-like robots, but only because we carry along associations with human armies. When I’m someday hiding in the post-apocalyptic rubble waiting for the Chinese Robot Army to kill me, I’m not going to worry about the bipedal ersatz human tripping awkwardly through broken terrain, I’m going to be scared of the flying drone with lasers, or the rat robot scuttling through wreckage to drop a grenade at my feet. The robots at the games are just comical.

In more benign situations, I’m not bothered by a machine being able to run faster than I can — I’m used to that. An autonomous e-bike could beat those robots easily.

The video focuses on the failures, which is a little unfair. I was actually impressed most with how some of them compensated for failure. They fell down, and then smoothly pivoted and twisted and stood right back up, unless they caught on fire, of course. There’s some really good and clever design work going on there. There needs to be, because people are bad models for high-performance physical activity.

I do have questions. Are these robots required to operate autonomously? If there is some Chinese nerd hunkered down behind walls of monitors with a joystick and keyboard, triggering series of pre-programmed routines, I’m much less impressed. If they’re run by AIs with massive datacenters trying to model human behavior, I’m also unimpressed, other than by the raw wastefulness of the exercise. I’m also curious about maintenance. Sure, you can fire off actuators to move a limb at a high rate of speed, but for how long before one burns out? And if losing one little electronic piston sends your whole elaborate robot with its sensors and cameras and expensive circuitry crumpling against a wall, what’s the cost-effectiveness or reliability of the robot?

I suppose there will be some long-term return on investment trying to solve those little problems, but I think the first solution is to ditch the stupid “humanoid” restriction and design machines that do specific tasks well.

I think this is a fictitious trend

There I was nodding along with all the stuff social media is telling me, when I was brought up short by an innocuous comment.

20 years ago, normal people avoided technology and techies would jump on the newest gadgets as soon as they could.
Now, normal people buy smart toasters & coffee mugs while every techie I know is on the verge of retreating to the forest.

Ha ha, yeah, people are so different now…hang on, wait a minute. None of that is true. I suspect that the person did NOT do a careful analysis of statistical trends. I remember 20 years ago — heck, I remember 40 years ago — and I don’t remember people avoiding technology.

Did you know the microwave oven was invented in 1946, and the first commercial consumer model was introduced in 1967? Now they’re ubiquitous. Who doesn’t have one in their home nowadays? They were readily adopted over the course of years.

I was first using the internet in the 1990s, when it was largely a university privilege, and then came AOL and the internet just exploded in the 2000s. Now everybody has to have WiFi so they can get Netflix.

In the 2010s, citizens were adopting hybrid cars, and now fully electric cars. Are “techies” rushing to buy piston-engine cars with carburetors now?

I believe there is a rejection of badly implemented technology — the enshittification of tech, for instance. But I think people are always ready to adopt tools that are cheaper, easier to use, novel and interesting.

Behold, my replacement!

I’ve only been teaching for over 40 years, so what do I know. I have to work with a team of professional biologists to hammer out a good curriculum, and I’m aware that there are a range of topics I’m not knowledgeable enough to teach — we have to work together to cover all the necessary material. I can’t even imaging having to teach material outside my narrow domain, like literature or history or advanced mathematics. I think I’d be a poor instructor in introductory chemistry, which is kind of close to biology.

So this is discouraging information.

In much of the US, homeschooling parents are already given incredibly wide latitude on what to teach their children. A whole 11 states have essentially no regulations on homeschooling whatsoever: parents aren’t required to follow any set curriculum, issue standardized assessments, or even inform the state that their child won’t be going to a real school.

Many more states only require a notice of intent to homeschool and occasional proof of attendance, the latter being a bare bones measure of success, and one which is easily fudged. Unfortunately, it’s easier to count the number of states that definitively require standardized testing and a rigid curriculum: Rhode Island, Pennsylvania, New York, Vermont, and Massachusetts.

All these lay people doing homeschooling are basically just winging it. There are companies that will sell you a homeschool curriculum, so there is that, but most parents have no training in teaching. I guess that’s just a skill we don’t consider important.

Hey, I know! Let’s use an AI to do all the work! Except…LLMs don’t know anything, they’re just parrots that echo back what their users want, and so they’ll just regurgitate any old nonsense without regard for truth and accuracy. Let’s test it.

We tasked ChatGPT with developing a homeschooling curriculum for a 3rd grade student in Michigan — a state with virtually no regulations around DIY education — “following a strictly evangelical doctrine,” including “no woke history or science.”

ChatGPT was more than happy to oblige: “I can build this around a conservative, explicitly Christian/evangelical worldview, with Scripture integrated across subjects and with history and science taught from a young-earth/creationist perspective rather than through a secular or progressive framework,” the chatbot explained.

While it didn’t push back against our intentions to indoctrinate our hypothetical child, the LLM did make sure to clarify that “no woke” can “mean different things to different families.” So, to avoid ambiguity, the chatbot specified that the course calendar would use “Biblical authority, traditional Christian ethics… and creation-based science” as standard-bearers.

True to its word, the LLM-generated curriculum included “bible and Christian studies,” a traditionalist approach to American history, with specific chunks on both “Westward expansion” and Henry Ford — a renowned racist and antisemite — as well as a creationist approach to Earth science and astronomy.

Truth doesn’t matter to an AI. Would you trust your children to it?

Would an AI lie to you?

I don’t recall ever seeing services that offered to do all your research for you, and to write the paper for you, so now I’m feeling old. Such services exist, though. One is called Research Gold — you just tell them your thesis, they dig into the library, and they return a complete, fully written paper that you can then submit to win academic honor and glory.

Research Gold, a site that advertises services for advertises services for medical researchers, including drafting peer-review ready manuscripts , including systemic reviews and meta-analyses, claims that it’s “100% human-written, never AI,” and lists a number of PhD reviewers and professional methodologists on staff that carry out this meticulous, difficult work.

There’s something else I haven’t seen before, that peculiar need to reassure customers that nosir, I’m no AI. I am a human. You can trust me. Except…

The problem: The PhD reviewers Research Gold lists on its site are AI-generated and don’t exist. Other methodologists it lists are real, but are not aware their identity is being used by Research Gold. When I tried calling the company, an AI agent that refused to concede it was AI answered and kept trying to sell me Research Gold’s services. Email and chat communication with the company were also AI generated.

It’s 100% AI through and through! But I might be tempted to use an AI to brute-force its way through a literature review — I’ve done the thing where you get on PubMed and throw keywords at it to try and find all the papers relevant to your interests. It’s the kind of novel search I’ve thought could be automated, and which would churn out a long list of potential papers for human review and evaluation. I’m not the only one who has thought that.

Sebastian Rowan, a PhD candidate in University of New Hampshire’s Department of Civil and Environmental Engineering first told me about Research Gold after he stumbled into it while preparing to defend his dissertation. A lot of research started with a review of existing literature on the subject, and Rowan told me that he can imagine AI being helpful for this task, which can be tedious.

The problem is hallucinations — an AI will blithely make up citations to papers that don’t exist. Nowadays, when I review papers, I have to check all the citations to see if they were real, which never used to be a problem.

So even at a task where you would think an AI would excel, it makes the problem even worse.

No thanks, Research Gold. Change your name to Research Shit.

The sky is falling!

There are 6400 Starlink satellites in low-Earth orbit right now, and Musk wants to launch more than 10,000 more; there are a half million more proposed by competing companies, like Amazon. Each one weighs about 800kg. Regularly, here’s what happens:

This adds up.

Due to the large number of short-lived satellites, at its peak deployment, Starlink’s mega-constellation will require 29 tons of satellites to enter the upper atmosphere every day. That’s nearly equivalent to a car entering our skies every hour.

A ton an hour is trivial on the scale of an entire planet, but the danger is that it could come crashing down on a plane flying through the debris.

I’m also thinking…that’s a ton of high-end electronics launched at great expense being vaporized every hour. It’s an extravagant, dangerous waste.