As I write this, nVidia is worth $2 trillion and Facebook is gaining about $1 billion in market capitalization per minute. A few minutes ago it was worth $1.275 trillion. Now it is worth $1.282 trillion.

The reason for these extreme valuations, as anyone who has been reading a newspaper over the past year knows, is AI. The term is everywhere. Every executive has to mention AI repeatedly in their earnings calls. Every startup founder is now an AI specialist, having been working on “an AI project” their entire life. Every venture capitalist is super excited to fund the next generation of AI technologies. Every acquisition can be justified by AI. Every major technology company has completely re-oriented its organization chart around new AI priorities. And every hiring or firing can be rationalized by AI: obviously, companies need more AI specialists, and fewer non-AI specialists.

Now that I have written the above paragraph, which took about two minutes, Facebook is worth $2 billion more.

Can we hold on just a minute?

First, the obvious: this is all a massive bubble that will end not just badly, but catastrophically. And it's both Elon Musk's and Sam Altman's fault. Musk's, because he constantly claims that artificial intelligence is the single greatest threat to humanity while working as hard as he can to make that threat a reality, and Altman's for releasing it onto the world far too early for his own personal gain, which, notably, has been substantial. Within the span of a year Altman has gone from mediocre startup CEO who used to run an early-stage investment fund with a bunch of goofy, invasive, sometimes illegal, and conflicted personal side-bets to humanity's technology savior.

Oops, Facebook just added another $1 billion.

Seriously, can we just pause? For a second?

AI has changed nothing. ChatGPT is an impressive chatbot, it's true. It writes well and answers a lot of questions well and it can even write some intermediate-level code. Its failings have been, at this point, well documented: hallucinations, wrong answers, and even disturbing indications of scheming and lying. But am I any more productive in 2024 thanks to AI than I was in 2023? No, I am not. I am less productive, because firstly, I do not use AI for a single task—no business task I need to complete can be accurately completed by an AI algorithm today—secondly, AI is being used to make life worse for everyone, which is actually the single greatest risk it poses in the short term, and thirdly, internet search has declined in quality because of technology executives panicking about AI. This fundamental point, that AI is an entropy accelerator, a chaos agent, and a weapon in the hands of malicious actors (and especially state actors) has completely escaped the notice of Wall Street. The fully-baked-in assumption that the future will depend on millions of nVidia chips grinding away at equations to AI-ify everything is as comically short-sighted and frankly, stupid, as the mathematician Liebniz assuming that he could sit at a table with his fellow mathematicians and pronounce, “Gentlemen, let us calculate,” to solve all of the world's problems in the late 1600s.

It is true that AI algorithms can generate incredible art and even photo-realistic videos, tasks that it will only get better at. But I do not make a living by generating art or photo-realistic videos depicting fantasy worlds, nor do I have any desire to. I write software that solves problems in the real world. In its current form, at this stage of development, AI is surprisingly inept at what I would like it to do. I can say this because I've tried using large language models in conjunction with practical code, and, in essence, it doesn't work.

But no one cares. Silicon Valley has abandoned the notion that companies should exist to solve problems individuals cannot solve on their own. The goal is simply money at any cost. Fake accounts (that destroy democracy), fake sales (that destroy price discovery), fake business models (that destroy the rule of law and housing stability), fake money itself (that destroys the environment and accelerates global warming)—these are what Silicon Valley calls "innovation" now. And in a valley where Greater Fool Theory is the predominant way that people make lots of money, the stampedes of fools chasing the reverberations of hype are getting so volatile and loud that they risk destabilizing the Earth itself.

The AI market bubble will end, because it is based on a false premise. The internet really did connect everyone. AI does not really work, nor do I think it will for another ten years at least, when it's possible—or even likely—that some key non-deterministic business tasks will be fully automated. But the collective delusion doesn't have to end soon, and before it does, we are facing some terrifying risks. AI won't make the world a better place, except for a few people like Vladimir Putin, Xi Jinping, Kim Jong Un, and various Ayatollahs, who seek to destroy order. Right now, that is AI's true strength. Silicon Valley and Wall Street would do well to take note, before they and their AI fantasies destroy us all.