OpenAI has literally solved on the order of 0.1% of open maths problems in three weeks. If the tech keeps ramping up, most of the entire library of open problems could be solved in a couple years. This guy is legitimately going to rope, isn't he?
>>17058904I remember watching the interview he had with the chief researcher at OpenAI back when o1 launched, he wanted to appear pro-tech and pro-progress, but you could tell it was all reluctant and he was seething behind the scenes.
>>17058904it's alright because solving open problems opens up new ones
>>17058931Lmao
It is the poor craftsman who blames his tools.AI tools will rise up to destroy us all, however.So, they are both right in their own ways.
There is an element of seething, sure. But no one has addressed his actual points, which is that when you solve the problem manually you have new insights that are useful not just for this specific problem but in general. Are there any new insights from that openai solution?
>>17058904>OpenAI has literally solvedThey used someone else's research in multiple cases.Also, LLMs only "solved" multiple problems by providing counterexamples. IMO this means there are whole classes of problems it might not ever be able to help with.
>>17059549Why do you think mathematicians or any one is owed any insight?
>>17059552You are retarded.
>>17059550>They used someone else's research in multiple casesLike literally every other researcher on the planet.The entirety of human knowledge (including math and science) owes virtually all of itself to previous discoveries made by other people, with the marginal gains in progress almost always being nearly entirely credited to the one or few people that added the single step forward. It goes without sayintg that OpenAI referenced at least a few of other people's breakthroughs, but this does not at all take way from the fact that AI itself solved Navier-Stokes, period.Had it been some dorky, shy, introverted prodigy alumni that all of you pseudo-intellectuals dream of being that had solved Navier-Stokes with even half the elegance or genius of OpenAI's model, they would be lauded as a genius among geniuses for decades, maybe even centuries to come.But no, its not good enough for you outwardly wholistic, yet inwardly self-centered luddites, that a frontier AI model did it. Not you. AI.No, it just *has* to be a genetically similar meat computer, not a silicone one that is far smarter and more knowledgable than you or I ever will be. To put it simply and in a way that you cannot refute: you didn't solve Navier-Stokes, and neither did stuck up Terrence Tao. No, it was AI, and there's nothing you can do to dim this miracle.You will be remembered as a modern day electricity skeptic (pic related)./thread
>>17059574P.S. reply in disagreement if you're a coping luddite lmao
>>17059010Not ones humans will be able to understand.
>>17059549Chat solved the problem, by hand, in an 88-hour token binge, and gained exactly 0 insight into the nature of the problem.
>>17059574>Like literally every other researcher on the planet.Lol no, they literally copied their paper before they made their findings public.
>>17059574I feel for the person who drew that, they must have witnessed someone being electrocuted, that's the posture they make
>>17059574>mental illness: the post
https://youtu.be/gXlfXirQF3A?si=Z_a-8rUXGXAUrbLVsong related
>>17059575>Makes an argument>>Btw if you try to argue against me you're a big doodoo headGenius.
>>17058904KWAB
>>17058904Hope so, im tired of seeing his ugly disgusting face.
>>17058904He loves llms he was one of the first big mathematicians i saw using it heavily. Also people who are highly skilled in their field can prompt information out of it in ways normies cant even follow along with, it doesn't surprise me that /sci/ cant understand this though
>>17058904It still can't produce a simple combinatorial formula for a 3x5xN binary matrix that has no monochromatic row in any direction. This took about 77 minutes of "compute" on an old laptop from 2002.
>>17058904applied physics doesn't care about your stupid theories.
>>17058904They should have AI take a crack at verifying IUTT.
>>17058904No it didn't. OpenAI is plagiarizing the work of Mathematicians.All CEOs and investors involved in AI must be shot in the head.
>>17066331I mostly agree with this, but for different reasons.
