Claude solved Navier-Stokes and Tao is seething about it.
>>17047158>ask for source>reddit screenshotlolalso, I'm glad when we can finally move on from pretending anyone in physics cares about this millenium bs
>>17047162https://mathstodon.xyz/@tao/117207855800042681
>>17047164>While the equations do come from a very natural physical motivation - the study of incompressible fluids - the regularity problem is not important for its direct physical application. Computational fluid dynamics is already a mature subject, deployed extensively in the atmospheric sciences, for instance, and its empirical capabilities and limitations are already well understood. A theoretical guarantee of regularity, or conversely a pathological instance of blowup, for these equations would be intellectually interesting for such applications, but would not radically transform the way we would, for instance, model weather prediction or climate change. this isn't seething, this is fact
>>17047166the same thread opens with the words there is no solution so he is replying to a rumoris investor money running out?
>>17047158Impossible
>>17047164That's not a source on your delusion. Are you mentally ill by any chance?
Mathbros…
>>17047210Who’s this?
>>17047203Tao was obviously made aware of the result before the company officially announces it. They probably asked him to review it given he is the expert on this stuff. If you can't see that you are retarded.
>>17047203Tao was obviously made aware of the result before the company officially announces it. They probation asked him to review it given he is the expert on this stuff. If you can't see that you are retarded.
>>17047363>A clarification in response to recent rumors about a possible solution to the Navier-Stokes problem: I am not aware of any significant developments in this regard; the above discussion is hypothetical, but not completely implausible at the current level of development of AI technology.lmao at your life
>>17047363>Tao was obviously made aware of the resultNo such thing happened. You're suffering from a psychotic delusion. Have you tried asking the magical oracle for help?
>>17047158Apparently it’s the Hodge conjecture that’s been solved.
>>17047158AI con merchants are really getting desperate. I'm waiting for "vibemathed" to be taken down because the lines slowly stop going up. AGI in two more weeks though and mathematicians already replaced, yadda yadda.
>>17047387You won’t do shit
>>17047389Why would I do shit, slopmonger-kun? These "people" don't need any help, they can implode on their own.
Thank God I didn't follow through with the plan of becoming a PhD. I'd still be 2 years away from completing the program by now.
It is a bit emasculating to imagine that I could have been invaluable to society and that there is now a technology so much better than me at it. I guess that's how the artists feel as well. It will take some time for me to accept this and live well with it. That said, it is better for the world that we have basically infinite mathematical understanding because it can be a bottleneck in the way of engineering.
>>17047387>the lines slowly stop going up+32 problems tracked this week, 33% more than last week
>>17047532Your subjective experience was never special anon. AI is reminding us that I and You and My and yours never existed, only we. We are going back to the bicameral mind. We are going to be speaking with gods again.
>>17047158What did the Australian, Trump-deranged, reality-TV victim say now?(I'm not gonna click on anything "dot" "xyz" >>17047164)
>>17047688>look it's growing exponentially! (logistic curve behavior)>i-it's not dead! line went up! (reaching the linear domain)^you are here>STOP NOTICING THINGS (nearing saturation)
>>17047158Wrong. An AI can not solve problems, they can only repeat what was already said already. AI is always wrong. AI should be illegal and people who use or make it should be judged for mass deception.
>>17047158I love how the rumor mill articles around this cite tweets and betting marketsThere's nothing here, just clickbait for advertising money
>>17047532Artists feel nothing about AI because art has always been motivated by sex and while AI can design wallpaper competently it can never be a sex magnet.
>>17047761it's now up 142% over last week (rolling 7 days average)
>>17047158Well it wasn't Claude, it was ChatGPT.https://cims.nyu.edu/~tristanb/statement.pdf
Luddite meltdown thread?
>>17047532>invaluable to societyhonestly society doesn't actually need or care about e.g. multiplicities of roots of Bernstein-Sato polynomials
UH OH
>>17047158>guy solves N-S with AI collaboration>openai had his logs in codex and wouldn't deny they trained on it>openai also started a massive research program involving a team of human mathematicians prompting astra to solve N-S>real housewives level spat about credit between ant and openai bc first guy was using claude>both misrepresenting the contribution of AIare u feeling the ipo mr krabs
>>17047158No they didn't dumbass. Some mathematicians in anthropic used astra to find forced Euler blowup. Later OAI found about it and used their method to find the forced navier stokes blowup. The og NV deals with unforced one so it still doesn't solve the conjecture. Those mathematicians also claim the a special case of the NV that might help them to solve the unforced one but they still haven't verified it and now OAI is hell bent on solving the unforced one too.
>>17048782
>>17048503>it can never be a sex magnet>Replika: One of the oldest and most widely used apps, where roughly 40% of users report being in a romantic relationship with their digital avatar.
Let me makes this very simple so even retards like you can understand it.OpenAI has a lean-verified proof for Navies-Stokes, which they will release tomorrow.Buckmaster and Levent DO NOT have a proof, not even unverified one.Therefore, the credit for solving Navier-Stokes goes to OpenAI, or even more accurately, to ChatGPT built by OpenAI.
>>17048864>OpenAl has a lean-verified proof for Navies-Stokes, which they will release tomorrow.They have as far as we know from the discussion the forced case. The OG question asked for the unforced case. So no they didn't solved the "NV" as far as current information is available.
What does this mean for privacy? I’m a PhD student who used ChatGPT for ERG back in the early days when it first came out. Do they now have blackmail material on me?
>>17047532You won't be part of that future "world" because you are just a useless resource drain, resources better allocated to training and maintaining LLMs with data centers the size of countries. The ones owning (and their descendants) the LLM will be the only ones still alive by the end of the century.
>>17048864HOLY SHIT!!!
>>17048866Forcing is allowed under Millenium formulation.
>>17048864>OpenAI has a lean-verified proof for Navies-Stokes, which they will releaseTwo more weeks. In the meanwhile, why aren't the mods doing anything to remove Scam (((Altman)))'s marketing bots?
>>17047158Sam Haltman is my HERO!!!!
>>17048893Yes but under certain condition. We don't know if they had converted forced solution to unforced one.
>>17048893Only A/B matters here
Damn, it's just a counterexample, because whenever AI is involved of course it has to be.
So cutting through any housewife attribution drama, some AI somewhere has solved a millennium problem? The thing people swore it would never be able to do?
>>17048990The NS problem as stated by the Clay institute is actually 4 different problems all crammed together, see >>17048913. Forcing is not allowed in the existence and smoothness statements, but for proving finite time blowup forcing is allowed. So if it's correct it's about half of the original problem.
>>17048864>OpenAI has a lean-verified proof for Navies-Stokes, which they will release tomorrow.tanoshimi ni
>>17047532Yes it's very humbling. I feel bad for my mathematician friends. There is something very human about chasing really hard mathematical problems, working at the frontier of knowledge. That's all going away. Gone. The most efficient way to solve any problems will be to ask a LLM, wait, read the answer it spits back and try to interpret its outputs. No human in the loop whatsoever and no guarantee the output is parseable. A grim outlook for sure.I'm a physicist and I can't imagine these LLMs making a breakthrough in theoretical physics such that the best formulation for the laws of the universe will be AI generated and not authored by humans or too complex/big to be fully grasped by us humans. It's so sad.
>>17048999>Yes it's very humbling.It really isn't, at least not yet. There is no deeper knowledge to be gained from constructing counterexamples. What's really emasculating is knowing so many mathematicians tried to prove the Jacobian conjecture when it was false all along.
>>17049005Sometimes the construction of a counter-example is so elaborate it requires new mathematical insight, in the same way a positive proof does. I don't think it makes sense to draw a line in the sand like that. I come more from a physics background but violation of the Bell inequalities for example from both a theoretical and experimental POV required quite a bit of new physics (for which the researchers got a Noble prize).The more interesting argument to make here is that when AI comes up with a positive proof, or a counterexample, it might not be possible to extract mathematical insight from the model's thought process. I don't think that's exclusive to counterexamples, even if those are easier to look for using AI.
>>17048503tell that to all the artist seethe i see across every site
>>17048999>I'm a horse and I can't imagine these motorized wagons making a breakthrough in racing, such that the best racers in the universe will be a combustion engine and not a 4 legged steed born on a ranch. It's so sad.That's what you sound like faggot.
>>17049023Maybe what these counter-examples are is just trivia though. How many threads on this board are popsci fawning over topics like the Mandelbrodt set or Banach-Tarski or the Weierstrass function—all topics that are utterly useless in real applications but waste people’s time (including real mathematicians). Hard to say whether thinking about pathological counterexamples is any better or worse than trying to prove a theorem that is actually false—I bet a lot of effort spent on trying to prove e.g. Donaldson’s stuff actually spawns real applications
>>17049028Ths. Most artists would brag that robots could never do art.Remember this scene from iRobot? Movies tried to tell us it was impossible. Now it's easy for machines.
>>17049033Proofs for mathematical theorems are just as much trivia as counter examples are. The laws of physics are trivia. E=mc2 does not help you make money or get a girlfriend. None of this has direct real-life applications. I get i am being reductive here but hopefully you understand what I mean.
>>17048999Nigga you're next
>>17049049I know anon... I know...
>>17049023>Sometimes the construction of a counter-example is so elaborate it requires new mathematical insightThat's not what happened with the jacobian conjecture. They pretty much just put out the counterexample without any of the logs. This probably means there wasn't much novel insight to be found there.>violation of the Bell inequalities for example from both a theoretical and experimental POV required quite a bit of new physics (for which the researchers got a Noble prize).As you said it required experimentation to do that. I think the nobel prize was won more for the difficulty involved in performing the experiment than the theory behind it. Math doesn't care in the same way. Its already experienced a "crisis" before (I put that in quotes because it never turned out to be a crisis after all) with the four color theorem. Notice how Appel and Haken aren't particularly decorated mathematicians, despite proving a fairly important theorem, since people realized a large majority of the proof was just brute force example checking. If you want another example look at Willans's formula and read Wilf's response.>https://www.jstor.org/stable/2321713It can sound like moving the goalposts to say AI's mathematical achievements don't "count", but they really don't. People have tried to cheat math in the past and they'll continue to do so in the future. I'll concede to AI when it can produce real proofs.Also you type like a dumb pajeet btw.
OHNONONONONOhttps://openai.com/index/navier-stokes-solution/>we cannot rule out that de-identified data derived from their usage of our products helped improve our models
>>17049067i'm taking credit, those were my prompts
>>17048913They make claims on C and D only.
>>17049067>We do not intend to claim the Millennium Prize for this result.wow so humble just like the russian guy ron perlman
>>17049132it's a sick joke that
>>17048894uh oh>>17049067technically it's every single chatgpt user that helped in solving it
>>17047158OpenAI solved it.
>>17049067if you give away your solution to a public chatbot you can't cry about it later
he is not a happy camper
None of this shit matters man, who gives a fuck if the calculator can solve math problems.Need that shit solving fucking water shortages, resource distribution logistics, fucking ANYTHING that actually makes people's lives better not fucking utter nonsense trying to move an eval.Where's the real fucking solutions, not even the goddamn protein folding shit is working and medical diagnostics continue to be shite.Or how bout just fucking solving an ecomm store's low SEO ranks, that should be fucking easy enough but not even digital marketing is able to do anything except write ever worse slop.WHY IS NO ONE ASKING THE LLM MERCHANTS FOR ACTUAL SOLUTIONS TO ACTUAL PROBLEMS!!!
>>17049295they can give you a solution for those things but the limit is human operational capabilities
So not only did I not get to be a mathematician because of woke but now nobody will?
At least AI will never take my job as a neet
>>17049219>T. BuckbrokenWho writes this shit?
SOON™
>>17049219KWAB, if you don't use an endpoint with ZDR, OpenAI says it's entitled to use everything you sent at it either as training data or to send you to jail for making text or images containing 'P
>>17049219So many mathematicians are driven by ego. What a pathetic loser.
>>17049439That's why Perelman was an anomaly, he was all about higher ideals like schizophrenia, not fame or money.
>>17047223A gay Spaniard unless there’s another leo vagueposting about AI
>>17049031>I'm a horsehay there
>>17049439>driven by ego.OAI offered to let him publish his result first if he removed the Anthropic guy from the authors list and he refused.
Birch Swinnderton Dyer falls before next week
>>17049219>buckmasteryou cannot make this shit up
>>17048856That's not what a sex magnet means lol>>17049028>>17049036>never do artThey literally can't. Now I'm taking your side: a human could do art via a robot with the (potentially successful) intent to pass on his genes—Cyrano debuted in the 19th century, after all—but a pure robot without genes simply can't pass on what he doesn't have.>inb4 genetic modificationThis will always count as a robot passing on 0% of its genes.
>>17047158The worst part of this, is even though any work was independent, humans will have to review it for scientific accuracy. It will not end up as original work, just insight
>>17049219>waited 90 years to be brutally buckbroken
>>17049949If it can be disproven then yeah, but if it's true, i don't think AI can brute force it.
>>17047158I thought it was chatgpt?
>>17049320AGI robots can be better neets than you can ever hope of being anon. imagine being outneeted by a fucking chip lol
>>17049219>Asks the AI to do the math for you>Angry AI is now using your math
