Maybe some of you have heard the claims that the Navier-Stokes Millenium Prize Problem has been solved. OpenAI claims that one of their "internal models" have solved it in the span of 88 hours. Unfortunately for me, my level in Mathematics is not good enough at this moment as to understand all the intricacities of the solution. But from what I understand, they found a solution that leads to vortices with infinite velocity but finite energy, which you cannot have in Physics. In other words, they managed to break the equations by brute-forcing a counterexample.
But that's not the most interesting part. The most interesting part are the worries of a few researchers who were working on that problem, that the model was trained on their drafts just very shortly before they intended to publish. Some people claim that they didn't have the original proof, but some proof of an Euler solution. The relationship between the two is beyond my current level of Mathematics.
OpenAI, allegedly, "heard rumors that two Millennium Prize problems had been resolved" and decided to test their model. I thought only Poincarre's conjecture had been solved so far, and Navier-Stokes would have been second. But anyway.
I'm posting part of a statement released by one of the researchers that have been working on this problem. You can find the full statement in the second link.
"I asked when the first prompt had been sent by them. This question was not answered directly by OpenAI for some time. Eventually it was agreed that it had been sent in the past few days, after information about our work had reached OpenAI.
I asked whether the model had been trained on, or had access to, our sessions in Codex, into which we had been putting all our drafts for the whole of this project. I was told the model did not look up user data. I asked again, about training, and I did not get an answer."
I think it's of utmost importance to keep people informed about the level of depravity that is infiltrating the sciences. I'm posting links to the OpenAI's statement, and Tristan Buckmaster's statement on this, who was part of the team working on this problem.
https://openai.com/index/navier-stokes-solution/
https://cims.nyu.edu/~tristanb/statement.pdf