OpenAI solves one of the biggest open math problems, HUGE!!!

[Legend]

We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics.

The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra.

The problem… pic.twitter.com/8zol3BPTL4

— OpenAI (@OpenAI) September 8, 2026

This is a very big deal. $1 million bounty for this problem.

[Legend]

pic.twitter.com/GsVdaqoJec

— whocareslol (@whocareslo4730) September 10, 2026

Also, more problems might be solved. This is crazy.

[kitler53]

is it confirmed solved or did AI just make up some shame and we can't prove it's wrong?

[Legend]

is it confirmed solved or did AI just make up some shame and we can't prove it's wrong?

A non ai math program has confirmed the AI made no logical mistakes, it is legit in that regard.

But it still needs to be independently verified and that will take time.

[Legend]

Twenty-five Fields Medal winners have published a joint declaration warning about what they see as a severe misalignment between AI companies and the mathematics community. pic.twitter.com/pFwHDX4Lq2

— Andrew Curran (@AndrewCurran_) September 11, 2026

What fudgy losers. Many math people have disliked computer assisted proofs for decades, but solutions are solutions!!! That's the point of math!!!

[darkknightkryta]

I was off-hand looking into this, since, whatever program the A.i. wrote could easily have been written by someone.  While neural nets are really good at finding patterns and inferencing data from said patterns, this didn't make sense to me.  Low and behold, scientists and mathematicians were working on the problem and Chatgpt (Open ai?) stole all their prompts.

[Legend]

I was off-hand looking into this, since, whatever program the A.i. wrote could easily have been written by someone.  While neural nets are really good at finding patterns and inferencing data from said patterns, this didn't make sense to me.  Low and behold, scientists and mathematicians were working on the problem and Chatgpt (Open ai?) stole all their prompts.

Yeah that's not actually true lol. Just one of those mathematicians being frustrated and making unfounded public accusations.

OpenAI was inspired because they heard a rumor that this problem was solved by Anthropic, so they wanted to see if chatgpt could solve it too. And then once chatgpt solved it they realized anthropic had only made progress but not actually solved it.

[darkknightkryta]

Yeah that's not actually true lol. Just one of those mathematicians being frustrated and making unfounded public accusations.

OpenAI was inspired because they heard a rumor that this problem was solved by Anthropic, so they wanted to see if chatgpt could solve it too. And then once chatgpt solved it they realized anthropic had only made progress but not actually solved it.

Still doesn't quell my thoughts, a scientist could have written a program to solve this.  Like, computers can't do anything they can't do.  What were scientists missing that the neural net found? And was it truly debunked?  

[Legend]

Still doesn't quell my thoughts, a scientist could have written a program to solve this.  Like, computers can't do anything they can't do.  What were scientists missing that the neural net found? And was it truly debunked? 

If a scientist could, then why didn't any? Cause even still it is only openai claiming this solution. No one else is claiming they have also solved it.

[darkknightkryta]

If a scientist could, then why didn't any? Cause even still it is only openai claiming this solution. No one else is claiming they have also solved it.

I know, but if it was an algorithm, what were they missing?  I also wouldn't dismiss the work mathematicians and scientists were working on it either.

[Legend]

I know, but if it was an algorithm, what were they missing?  I also wouldn't dismiss the work mathematicians and scientists were working on it either.

I guess in time things will be clear.

If this was a one off because humans were actually the source of the intelligence, no more breakthroughs will be coming. Or if it was indeed ~"hey chatgpt solve this, make no mistakes" then we will have plenty more announcements in the future.

The rumor is that there are plenty more but nothing is official till they make it official.

[the-pi-guy]

And was it truly debunked? 

From what I understand, the result hasn't been thoroughly verified and it'll take a lot of time. Still need to validate that the logic matches the original problem and that the original system is correct. 

I know, but if it was an algorithm, what were they missing?  I also wouldn't dismiss the work mathematicians and scientists were working on it either.

A big thing here is that they're running 10,000 models. 

This is basically throwing everything at the wall and seeing what sticks. 

AI technically has more memory than a person can have.