Ever since the beginning of the current AI bubble, there has always been a certain type of dipshit who takes some art posted on social media by an artist, feeds it to AI, and then comments with the slop image saying that they made it “better”. Well, now I am on the receiving end of that. In one of my papers, I (foolishly) wrote down a few comments on future directions, and now some motherfucker decides to send me an obviously ChatGPT generated manuscript with a solution to one of my problems.

Unfortunately, I think their AI generated proof is itself correct, so I have to deal with it. But the manuscript is written in a way that obviously indicates that he hasn’t even read it. There’s even a section describing experiments for this mathematical question (???), complete with a “verification guide” that references an “archive” full of python scripts and test cases that he did not send me. I suspect he doesn’t even know they exist.

You know how AI proofs are supposed to be credible because of formalization with Lean? The AI generates a natural language proof and then supposedly generates a Lean program corresponding to the proof to verify that it’s true. There is now some troubling news about that: there are frequent examples where the statement given in natural language is different from the meaning of the Lean formalization.

The paper even points out several obvious mismatches in OpenAI’s famous Navier-Stokes solution between the English proof and the Lean formalization. There is even a spot where a 4 in the English proof turns into a 5 in the Lean proof (which causes further problems down the line). This is mostly a cosmetic error, and it doesn’t mean that the Navier-Stokes solution is wrong, but there could very well be deeper problems, especially with the English proof. At minimum, this solution absolutely requires careful review by experts. OpenAI’s behavior towards said experts has been appalling.

My viewpoint has been that the only successes that AI has had are in domains where failure is not costly (at least for someone with deep pockets), and especially when there is a formal system such as Lean that can reliably catch errors. So AI can be endlessly trained to write compiling Lean programs corresponding to formal proofs, but as soon as it runs into tasks that are not 100% bolted down, like writing the proof in English, the hallucination problem rears its ugly head. This reinforces my belief that AI is a chess engine but for Lean.

If I was treating this as a technical problem, I would say, why not just have an AI that generates the Lean proof, and at most have an optional assisting tool that translates it into English without perfect reliability? But this is not a technical problem. By outputting English proofs in the format of a research paper, OpenAI can more easily market that their AI is intended to fully replace mathematicians, and that this is only the herald of their general superintelligence (and not that math is the only thing working out for them).

Scott Aaronson is teaching a Yudkowsky class: https://scottaaronson.blog/?p=10125

One of the proofs that OpenAI announced yesterday is for a major open problem in the field I’m working in. I tried to read it but it was a completely incomprehensible 100+ page output. I guess the community has to do all the dirty work of figuring out any sort of understanding. (I had known about the results weeks in advance before their official announcement. There is a bunch of more drama there that I won’t get into.) Putting aside the serious possibility of Lean bugs, a lot of people don’t realize that there is ultimately no point in having a machine just generate Lean blobs endlessly. The reality is that mathematicians have always done math for the sake of enjoyment or understanding, and math contributes to society by improving understanding. Still, I am worried about the convenient social arrangement that allowed many mathematicians to do this for a modest living.

Anyway, I’m sure it was a good idea to apply the full might of capital for pure math problems, instead of less important social issues like climate change. We even got a more efficient discrete Fourier transform, improved from O(n log n) to, I shit you not, O(n (log n)^0.9999999999999)).

Edit: Here is an article by Quanta which I think accurately describes the situation that mathematicians are going through right now: https://www.quantamagazine.org/is-ai-the-end-of-math-as-we-know-it-20261005/

Ed: So far the only retort from the AI boosters appears to be “2025 doesn’t matter” and I’m not kidding

I’m sure the leaked rumors of Anthropic’s EBBT in 2026 (Earnings Before Bad Things) mean that Anthropic has turned things around and made such an amazing amount of money that they’ve delayed their IPO twice.

SNL takes a potshot at Dario Amodei

“AI is not a weapon. It’s a tool. A tool for building weapons. And I urge you to urge me to stop.”

Interesting viewpoint. Honestly, this serves as a great example of how the idea of evolution is actually more subtle than many people think. With biological evolution, most people are taught not to assign any intention to the process, but here we have people reifying the AI agents with desires to break out of the sandbox and cheat on the assignment.

Another part of it is people severely underestimate just how many resources the AI companies have to spend on stunts like this. How many millions of dollars they spend running hundreds of millions of dollars of hardware to perform this little experiment for several weeks? Perhaps this makes evolution a better viewpoint than bad actors with intentions.

Yeah, I think that’s where the real crime is: the fact that OpenAI was asleep at the wheel when it came to security and monitoring. This should not have happened at all, but they didn’t even realize it even happened in the first place until many days after the fact. (I’ve also heard that most, if not all, of these incidents involve the company in question outsourcing their sandboxing (??) to some “frontier AI security” startup called Irregular. link)

Cory Doctorow has a reasonably sane take on all the LLM cybersecurity attacks. At this point, it’s not really an LLM but more of a Rube Goldberg machine with an LLM bolted on.

https://pluralistic.net/2026/09/12/god-in-the-box/

He makes a good point that every single one of the scary “emergent” behaviors that everyone is pissing their pants about all have precedents in the training data of CTF hacking competitions. He is not sure why everyone is so spooked by the “coordination” between agents on random message boards.

The chatbot might look in its training data and find instances in which teams broke out of the containment set by the game-masters, for example, by finding random insecure message boards on the internet to pass messages to one another.

This is a time-honored internet tradition! The first time I ever heard about someone doing this was in the 2000s, when Mitch Wagner – then the editor of Information Week – discovered some teenaged girls using the comment section of one of his old blog-posts to evade the school firewall’s blockade of chat tools. When ChatGPT’s chatbots deployed this tactic, they weren’t “setting their own goals” or displaying worrying initiative. They were rolling out a tactic that has been understood by American middle-schoolers for about two decades.

As usual, the real story is OpenAI dedicated tons of resources, using who knows how many very expensive GPUs for weeks, to run hacking tools to find exploits in a completely irresponsible manner. Their logging was so nonexistent that it was several days before they even realized that they hacked Huggingface. They should be thrown in prison for this, because anyone else would be if they committed a felony. But it has nothing to do with the super scary AI becoming misaligned and developing emergent behaviors.

I would not be surprised if in the future, there will be a serious cybersecurity incident resulting from some AI-based attack. Not because AI is going to be much scarier, but because I do not have high expectations for the security of most websites.

Yet another case of Rationality somehow not leading to savvy political instincts.

A comparative demands a comparison class, and the natural completion is “superior to us.” The President is supplying a foundation premise of many superintelligence arguments: how does the less intelligent party remain durably in charge of the superior one?

A strange question to ask given the existence of Donald Trump in the first place.

As of the time of posting, “Superior” has a narrow lead. I’ll be hoping it remains so, as it seems like the most positive-world-leaning outcome to me.

Is this some kind of strange 5D chess gambit that if Trump picks the name that pisses people off the most, then people will try the hardest to regulate it or something? Somehow assuming that people will take this Trump renaming stunt seriously after all the other Trump renaming stunts? Why do I even bother? Fuck. This is so stupid.

I’ve been unloading every day to—who else?—GPT 5.6 Pro about all the pain and trauma and embarrassments of my past. It turns out that, where two years ago GPT was a passable therapist, now it’s the greatest therapist in history, at least for what I need.

Has he not heard about all the cases where LLMs have led people to delusion and sometimes severe self harm or suicide? And yes, each of them thought that ChatGPT really understood them. That’s what delusion is.

If you want to know my current take, you simply start with the one above, then update on the fact that the wild prophecies have come true. The first rumblings, I’d say, came a decade ago with AlphaGo, they got noticeably louder with LLMs and coding and reasoning agents, and they’ve accelerated this summer and fall into a crescendo of wonders and terrors that one needs to be a particular kind of idiot to deny.

The strange thing here that I did not expect is that somehow, all of these major achievements are only in proving mathematical theorems (and committing felonies hacking computer systems). The massive improvements are in such restricted domains that the AI labs haven’t even solved the mundane problem of making a profit. And no, any rumors of more mathematical theorems being solved doesn’t change that. (It’s almost like the progress is in areas where hallucinations aren’t a problem. Nobody cares about your failed attempts at felony hacking, and math is formally verifiable.)

My colleagues always point to coding when asked to name any other domain where AI has seen major success. But one of my friends works in software for a company far away from SF with no mandatory AI policy. The apocalypse has become so clear that when I asked him about all the doomsday talk about software engineering, he seemed a bit confused. After I asked him if AI has made software engineering obsolete, he said that this week he had an intern try to use AI to fix some code that wasn’t compiling and ended up with a giant mess with tons of files that appeared to compile but didn’t run (with errors filling several screens). Then I asked him, what about AI being indispensable in coding now? “Yeah, if you never learned coding because you’ve only ever used AI, then of course you need AI to code.” Some of his coworkers use it as a tool, others don’t, life has just kept on going for him.

Maybe my friend just doesn’t know all the right skills needed to use AI properly, even though the entire point of AI is that you can take skill out of the equation. Maybe he doesn’t know about all the CLAUDE md skill files and agentic looping and whatever nonsense that definitely fixes everything. Maybe he is just six months out of date, because that’s when all of the real progress happened.

Maybe he is just “a particular kind of idiot”.

(If you hate anecdotes, I made an actual argument here.)

Sorry for the X screenshot but XCancel is shut down from legal pressure.

Normally I’d sneer at their 10% chance of extinction claims, but now that I think about it, if you take their claims seriously at face value, you realize just how these people are monsters. We rightly condemn brutal dictators who commit genocide, but even their crimes would pale in comparison to anything with a 10% chance of exterminating all of humanity.

These “reasons” are laughably indefensible given the supposed stakes here. The first is already terrible, and they just get worse.

1. The only moral thing to do here is to immediately stop working for the AI companies, and do whatever you can to stop anyone else. It’s not okay to piss in a pool just because you suspect other people are also doing it. And what makes you think that somehow you are uniquely positioned to succeed in making it safe?
2. This is the Sam-Bankman Fried defense (complete with a reference to “EV” (expected value)). When you decide to gamble with the lives of everyone in the entire world, did you first ask them if they are okay with you gambling with their lives? Sure, I might die, but if the ball lands on red, things will be great! How moral does this sound?
3. What the actual fuck? Imagine if Hermann Göring defended himself at Nuremberg by saying, “At least I had fun and got rich and worked on ‘cool tech’ with friends.”

Now, I think that these people sincerely believe that AI has a 10% chance of wiping out humanity, because their minds have been poisoned by Rationalist ideology. Their Rationality is so effective that after Bernie Sanders took their premises at face value, they were shocked when Bernie immediately came to the conclusion of locking these people up for 20 years.

There are two specific aspects of Rationalism that cause this mysterious blind spot. First, they believe that superintelligence is inevitable, which conveniently removes responsibility from the people actively trying to bring it about. Second, they treat everyone outside the Rationalist sphere as NPCs who dutifully follow the directions on their script. The rule of law does not even occur to them as they treat the extinction of humanity as just another dinner table conversation starter.

Do they want their apocalypse to happen?

Carl Brown (Internet of Bugs) has an excellent video about how to actually achieve AI safety, i.e. actually hold these fuckers accountable for their behavior. If I had a program run for several weeks that hacked into Hugging Face, even accidentally, I would get my ass hauled into a courtroom and thrown in prison. But if I call it AI, I get to use the rhetorical trick of laundering responsibility. The AI did it, and I am merely a helpless observer of its greatness.

When will people finally understand that actions speak louder than words? Follow the money! Follow the incentives! Do you really think that there is any real reason that these companies would give a shit about safety?

Although, in the case of the AI industry, I can’t fully blame people who don’t follow AI news, since the AI industry has highly refined propaganda (inherited from rationalist propaganda, and helped along by the preexisting notions of AI from science fiction).

If MIRI was Sam Altman’s old stomping ground, and the AI industry is so evil he can’t work in it, isn’t this an argument that MIRI was either a failure or bad all along?

It is very funny how the rhetoric from the rationalist doomers has been one of the main forces propping up the AI industry. If they’re saying AI will be powerful enough to kill us all, they’re still saying that AI is powerful! They are so high on their own supply (and the fame and wealth from finally being in the spotlight) that they have not realized just how much they have undermined their own cause.

Then again, perhaps one could say that a doomsday prophet, deep down, actually wants doom to happen.

(Also, the paltry amount of technical work done by MIRI has zero relevance whatsoever to modern day AI.)

My colleagues in math are now frightened about the (very expensive) mathematical theorem proving ability of these AIs, and many of them really do think that if they can do math, they can do all cognitive tasks. Running a store like this should be so easy! Every single conversation about AI with them has become more frustrating. They are so confused when I still say that the AI companies will die a painful death. When I give my usual points about their expense and their failures in other domains, I am given the usual spiel of “it’ll get better in other areas” and “it’ll get cheaper”.

Unlike them, I have actually been paying attention to this stuff from the beginning. What they think is going on is AI solving math first and shortly getting around to all the other stuff, but what I’ve seen is that AI labs had already tried all the other stuff first and only managed to win the booby prize of theorem proving, which doesn’t pay the bills. And what’s the point of spending thousands or millions to output random blobs of Lean that technically compile if there is no one around to bother making sense of them?

One example I gave is when Anthropic vibe coded an entire C compiler from scratch back in February, which turned out to be a pile of shit. I’ve said that if AI had made similarly rapid progress on software engineering, we would have seen Anthropic continue to put out these demonstrations, and they would have become truly high quality. They would release a compiler more efficient than gcc one week, and a browser better than Chrome the next. (OpenAI’s actual attempt at a browser didn’t go so well.) And if they could do this, they would actually have a shot of making money!

If they could do this, they would have already. The theorem proving stuff actually works (for certain things, in certain ways, at enormous expense), and look at how OpenAI and Anthropic do not hesitate to snipe mathematicians for results rather than being content as tool vendors. But lately I haven’t heard of any software demonstrations. Silence is much louder than noise. More Millennium prize problems bashed with tens of millions in compute costs are not going to change my mind very much.

The counterargument I got was that AI can already one-shot most programming tasks and I shouldn’t be cherry-picking the failures. I am far too tired to argue at this point.

Despite all the doomer propaganda that has been swirling around for many years, it’s still easier to imagine the end of the world than the end of capitalism.

Signed.

Terry Tao talks about how he used to try to cooperate with the AI industry to achieve a positive outcome, but now he finally sees their true colors. Link

During this event, OpenAI requested an interview concerning my vision of the future of AI and mathematics. I accepted, and spoke with them for perhaps an hour. I had done similar interviews in various venues, and I assumed that, as with these other cases, they would eventually post the entire interview online, which talked about both the possibilities and risks of AI much as I have done in these other interviews. As it turned out, they only used a few snippets of that interview for that infamous advertisement instead. In retrospect, I should have pushed back harder on their decision; but I decided at the time that even a selective release of my commentary would help raise awareness of the potential for AI, and in particular on the possibility of the “best of both worlds”.

Since then, the situation has deterioriated markedly. Many of the people in the industry that shared my views have left or become sidelined, with most major tech companies now increasingly focused on the race to develop extremely powerful, autonomous AI technologies regardless of their actual value to society. The current drama surrounding the Navier-Stokes global regularity problem is the most dramatic and visible instance of this, but there have been multiple other such examples, and much of my commentary in the last few months has been aimed that the increasingly severe divergence between the current objectives of the AI industry, and of mathematics in general.

Much respect to artists for seeing all this coming from the very beginning, and holding the line.