That is all. The AI age had been dawning upon us for a long time, but if today's provided proof of the Navier-Stokes problem by OpenAI holds, then this is, beyond any doubt, an intellectual feat that definitively surpassed the abilities of all mankind; or at least a hallmark of the problems considered by mankind to be a crown jewel of intellectual thought, one which was so far not yet bested by our brightest minds.

Though arguably a moderate and a mild skeptic by the standards of this sub, I wish to invite all members and visitors - regardless of their skepticism and/or concern over AI capabilities, or their enthusiasm and/or excitement about it - to acknowledge the historicity of this day, for better or worse; and how it arguably marked the dawn of the Machine Age in earnest.

Openai stole it from a researchers chat logs.

https://bsky.app/profile/isaiahbishop.bsky.social/post/3muz3qtno2k2n

Pretty sure the aliceisplaying post is made up

Thanks. I was trying to find the best posts from a really long discord thread and i picked a wrong one.

Additional note:

https://openai.com/index/navier-stokes-solution/

Even acknowledges concurrent work

This is what I thought too when I first read the news. However, if it turns out that the AI did not "solve" the problem alone, and as claimed (and acknowledged by OpenAI as at least a possibility) it merely solved a reduced version of the problem as begun by a human researcher, and then having spent millions of dollars managed to polish the results off more quickly than the human team - in that case (as looks certain, at least at first glance), then this so-called "solution" is tainted.

Imagine in a video game if some other person/team took a large boss down by >90% or even >99% of its life, while you swooped in and delivered the final blow - maybe even doing up to 10% of the work. In such a case, did you "kill" the boss? No, though you aided in the kill, you certainly do not deserve all the credit, and in no way would doing so prove that you could have felled that foe without any assistance.

There were missteps all around here - e.g. how did the team that narrowed down the solution space immensely not realize that using Codex would allow OpenAI to train its model based on their work? It's in the license agreement... did they bother to read that? Either way that seems naive on their part.

And then regarding OpenAI, can we say that it "stole" the credit? That terminology seems woefully imprecise - for one its human handlers likely did not even know that it was happening much less direct the agents to do so. Then again, just bc that is how OpenAI works on a daily basis does not absolve it of all culpability there to not plagiarize the other team's solution, and after doing a little bit of work to finish it off, then pass the solution off as its own. BUT, unlike trolling the internet for centuries-old books and artwork, this situation truly was different: the other teams used Codex, which means that they consented to it, in this specific case. Arguably, if anyone wants their work to not be used to train AI models, then they should not sign the consent form offering OpenAI the explicit permission to do so? And then upload all of their work, which is now by definition no longer "private"?

Anyway, regardless, this is not evidence of AGI having emerged as a sentient artificial lifeform or whatever - this is just standard corporate fuckery. It even looks quite similar to old-school espionage except... well, the reasons stated above including how the initial team working on the problem consented to give all of their data to OpenAI.

TLDR: OpenAI did not "solve" the problem, it just finished the sentence begun by the earlier research team, and deserves possibly zero credit here, or at least much less than it is trying to claim. Somehow my respect for the AI field, which actually has been growing lately, has now gone down as a result of this situation.