All of this text was written by a human, not even proof read or grammer checked by AI.
Recently an internal OpenAI model solved the Navier-Stokes Millennium Prize Problem, a very hard open mathematics problem, that humans could not solve since more than 100 years. There is huge controversy as of whether OpenAI "stole" some work of mathematicans Tristan Buckmaster and Levent Alpöge that were also using ChatGPT trying to solve the same problem. Even if that is the case and OpenAI stole some of that insight, why didn't Buckmaster and Alpöge solve the millenium problem before OpenAI, when they were aparrently so close? It doesn't really matter who solved the millenium problem anyway; both parties consisted of mathematicans that heavily used AI to solve a math problem and in the end it seems as OpenAI had access to a much more powerful internal model and ended up solving an Millennium Prize Problem (The prompt was probably "Solve Navier-Stokes and do not stop until you are finished")
I don't know all the details of that mess and if things were stolen, that is not good, but in the end what is important: Mathematicans plus strong AI lead to a Millennium Prize Problem being solved.
Furthermore, there are rumors that OpenAI and Anthropic made considerable progress with other Millennium Prize Problems, maybe even having solved them and holding back because they don't want to piss of mathematicans any further. What a bunch of weaklings (both sides).
In my humble opinion those little narcissitic math nerds are overreacting a bit. 25 Fields medallists signed a letter that condems the current use of AI to solve those math problems. Their claim is basically that it destroys the mathematical community and their work by grabbing the solution but not making any progress on the way, because apparently not only the solution is important, but also the way to the solution and what has been learned along.
I mean there is some merit to that criticism: Oftentimes, working on a problem forces you to solve sub problems that lead to completely new sub fields and insights. The path to reach the goal is sometimes more important than the goal itself.
I get that and I think part of this criticism is correct.
But honestly, the AI probably also made those detours and partial insights on it's way to finding the solution to Naiver-Stokes. Why can't the mathematicans just follow it's trail and update their knowledge by ingesting those sub fields?
The letter that the 25 fields medallist signed claims that AI mostly causese the following problems:
- math students and ideas develop through slow human processes (problem solving, talks, conferences, writeup of theses) that answers created by AI completely bypass. Solution: maybe those slow processes are not needed anymore? In fact if they are needed, they will still be done by humans until computers can do them better (and they will)
- results announced in a rush skip proper writeups and citation, raising attribution and plagiarism issues Solution: fixing citation issues is not a real problem, jesus christ. Probably AI cites better than most humans
- ideas nobody humanly integrates into the field of mathematics never become part of the field Solution: Why shouldn't AI be able to use it's own ideas? That claim really doesn't make any sense. At the end there is some paper with a lean proof and some new sub fields and this output will be part of the training data of the next model
I am not going to claim that they don't have points, but those are not real issues that cannot be solved by halting AI.
Their main point is simply: AI just proofs a theorem without doing all the foundational work, which in turn is not beneficial for mathematics in general. But if this is really a problem and if only humans can do this foundational work, then again humans will be paid to do exactly that work and integrate and understand those AI proofs. If there is a problem to be solved there, it will be paid labour and it will be solved.
It seems to me more that the little math nerds are just being salty that their little brains cannot work so fast and efficent and their main line of critcism is: "It's too good for me, please slow down"
I get it, it must suck that AI 2 years ago couldn't basically find out what number is larger, 9.11 or 9.9 and now it Millennium Prize Problems.
They were so sure that their field of work is the pinnacle of achievment and "nobody" could do their work, now reality just hits really hard.
To grown up way to react to that is similar to us small brained low IQ programmers 10 months ago in November 2025, when Anthropic basically fundamentally changed software engineering forever with the realease of Opus 4.5.
No one gave a fuck when our "programming art" was utterly altered. AI is writing code much better and more beautiful than the best programmer nowadays.
Plenty of us considered software engineering some kind of art. Coding was creating art.
Did the society of computer scientist and the Turing Award winners in computer science publish a similar letter as the fields medallist virgins?
I have a masters in computer science and I am programming essentialy since 20 years, Python, NodeJS, C whatever.
I have maybe 15 open source repositories with maybe 4000 github stars, I worked for free and this AI is partially trained on my code (hence why it still sometimes sucks)...
Nobody gave a fuck when my art was destroyed.
Writing beautiful code is awesome I get it, but the fucking reason I went into computer science is to basically automate things and to find a way to make money without spending my time. Passive income baby. I like to fucking program and I love solving problems, but coding was always a tool, a means to and end.
So is mathematics, it's a tool to advance sciences and to improve our life.
Let me be very blunt: I don't give a shit what virgin Fields medallist think in their narcissitic heads.
I want science to advance (in a safe way) so that real problems of real humans are being solved: Disease, poverty, war.
Right now there are still millions of people working jobs that absolutely no one wants to do, such as mining, logistics, cleaning, truck driving, fast food, delivery, taxi drivers, farming, fishing, and so on
We have humans dying from cancer, viral diseases and we have planes crashes, car accidents, traffic congestion, pollution and horrible diseases.
Mathematics needs to be advanced, so that other fields can advance.
I absolutely give zero fucks about the feelings of virgin nerds. I get it, now they have no more reasons that feeds their superiority complex, welcome to reality.
What if AI solves cancer (I know this is probably too simplistic) and then cancer researches state "we need to ban the cancer cure because the way to the cancer cure is more important than the cure itself"
what a bunch of morons that would be
yes it must suck for a lot of scientists that it is us mathematicans, computer scientists and ML engineers (who are essentially applied mathematicans) who give birth to an entity that is more intelligent in EVERY field than the best humans.
If we can do it in a safe way, a super intelligence is the best and most important invention ever done by a human.
The creators of this machine do not understand mathematics, but the machine itself understands it.