Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Navier-Stokes – Tristan Buckmaster [pdf]

cims.nyu.edu

Post reply on HN
Print view View original

831–840 of 869 posts

Re: Navier-Stokes – Tristan Buckmaster [pdf]

#831

Post by OceanSeth »

tristanj wrote: post #2

Full mastodon post: https://mathstodon.xyz/@tristanbuckmaster@mastodon.social/11... Mathematical explanation by Terrance Tao: https://mathstodon.xyz/@tao/117233527638291447 It seems there is much background drama behind this, and this is what I've pieced together of what happened: Over the past year, Buckmaster and Alpöge have been using AI to work on fluid dynamics maths problems. Alpöge works at Anthropic, which wi…

This reminds me of a conversation I had with an engineer at google when I was angrily saying "it's not end to end encryption if you get my emails and can train your models on them" to which he said "we try not to do that". What a statement!

Re: Navier-Stokes – Tristan Buckmaster [pdf]

#832

Post by lukewarm707 »

fc417fc802 wrote: post #735
za_creature wrote: post #616

Earlier quoted context omitted.

Many do indeed hold the position that all LLM output is uncopyrightable plagiarism. They're probably right, but there's an even stronger argument here: Science papers of a phd level must contain: 1. one or more novel insights 2. a long list of citations to contextualize them and 3. some work to prove that the insights are in fact meaningful --- In this context, consider a prompt based diffusion model which, when aske…

I neither agree nor disagree that all LLM outputs are plagiarism. I merely objected that the line of argument engaged in was specious given the context. As to your stronger argument. You only cite prior novel insights that you're actively building off of and that (approximately speaking) fall outside of the status quo. You don't for example cite leibniz or newton despite your paper making heavy use of calculus. So is…

evidence that openai trained on the data: they would have denied it if they didn't train on it.

did the proof build on the insights:

the influence of an individual text in the training data is deeply weighted by quality, relevance, etc. a high quality proof in advanced mathematics written by a codex user is going to get boosted to the max.

the model is post-trained on prompt material. that is again going to boost it.

the prompt will boost this material specifically. perhaps they even rammed dense maths in particular into the model in post training.

anecdotally i have been able to get near-verbatim copies of original material out of models at inference. the type of work that buckmaster and alpoge fed into openai feels like the exact type of concept that would cause an "aha!" or "but what if?" in chain of thought. in fact i would bet that their work is in the logs.

the likes of astra and fable are thought to be up to 10T parameters in size. i consider it highly plausible that a semantic representation of the euler proof could be pulled out of the model weights in good shape.

Re: Navier-Stokes – Tristan Buckmaster [pdf]

#833

Post by lukewarm707 »

fc417fc802 wrote: post #729
Tanjreeve wrote: post #675

Earlier quoted context omitted.

That's exactly the argument of the people calling it plagiarism machines. No-one ever really did refute it there was just a bunch of settlements for elite institutions so they weren't left empty handed like the various small time creators/authors etc were. I think the bigger issue here is this feels like some PR smoothing happening that after all the work that went into "it's safe to use for enterprises" now we have…

> now we have what looks like openAI using private user data to scoop novel research Is there any actual evidence of that? All I've seen so far are empty accusations because "it would be in their interests" or whatever. Personally I'm inclined to believe that they honor their terms until it's demonstrated otherwise.

their terms grant them an irrevocable license to your data unless you specifically opt out of it.

Re: Navier-Stokes – Tristan Buckmaster [pdf]

#834

Post by davidguetta »

mayakacz wrote: post #12

I'm not one to comment often but this really pisses me off. OpenAI looked at user data, stole world class researchers' work, and then tried to threaten those researchers to do what would make their corporation profit (which they would anyways!). Imagine you have been working on a terribly difficult math problem for a decade. This is a result you have spent years on, and what you will likely be remembered for. And to…

how much of the "codex discussion" was actually ideas also generated by open ai ? this is a bit the elephant in the room

Re: Navier-Stokes – Tristan Buckmaster [pdf]

#835

Post by burkaman »

corford wrote: post #785

One of the best takedowns of samas/OAIs spin on this I've seen is this post on X: https://xcancel.com/BetterCallMedhi/status/20974723362578637...

Who is this guy? Even if I had enough mathematical background to understand the arguments I'm not sure I could follow this post. Punctuation exists for practical reasons, it's not just an aesthetic choice.

Re: Navier-Stokes – Tristan Buckmaster [pdf]

#836

Post by westurner »

westurner wrote: post #395

"OpenAI has solved the Navier-Stokes Millennium problem using $15m of AI effort" https://www.newscientist.com/article/2588063-openai-has-solv... $15m in tokens; but what about labor? What about compressible fluids?

"Blowup for the Euler equations with smooth forcing" (2026) https://cims.nyu.edu/~tristanb/euler.pdf

"Blowup for the Boussinesq equations with smooth forcing" (2026) https://cims.nyu.edu/~tristanb/boussinesq.pdf

"Extending the Córdoba–Martínez-Zoroa IPM Blow-Up to Uniformly Spacetime Smooth Forcing" (2026) https://cims.nyu.edu/~tristanb/ipm.pdf

"Finite Time Blowup for the Euler equation" (2026) https://cdn.openai.com/pdf/315b36cd-ec98-4023-8342-93345194e...

Lean proof: openai/NavierStokesAndEuler: https://github.com/openai/NavierStokesAndEuler

Re: Navier-Stokes – Tristan Buckmaster [pdf]

#837

Post by westurner »

westurner wrote: post #836
westurner wrote: post #395

"OpenAI has solved the Navier-Stokes Millennium problem using $15m of AI effort" https://www.newscientist.com/article/2588063-openai-has-solv... $15m in tokens; but what about labor? What about compressible fluids?

"Blowup for the Euler equations with smooth forcing" (2026) https://cims.nyu.edu/~tristanb/euler.pdf "Blowup for the Boussinesq equations with smooth forcing" (2026) https://cims.nyu.edu/~tristanb/boussinesq.pdf "Extending the Córdoba–Martínez-Zoroa IPM Blow-Up to Uniformly Spacetime Smooth Forcing" (2026) https://cims.nyu.edu/~tristanb/ipm.pdf "Finite Time Blowup for the Euler equation" (2026) https://cdn.openai.com…

Is everything in NavierStokesAndEuler/NavierStokes [1] precedent to the derivations?

How to tree shake Lean4? (Edit: `lake shake`,)

[1] https://github.com/openai/NavierStokesAndEuler/tree/main/Nav...

Re: Navier-Stokes – Tristan Buckmaster [pdf]

#838

Post by burkaman »

naniel wrote: post #364

OpenAI's release explicitly says No. But then also caveats that with "we cannot rule out that de-identified data derived from their usage of our products" impacted things. What's most striking to me, and what may or may not be true, is the "we cannot rule out" bit. "We (the researchers and the agents) did not see any of their work through any means until they released it publicly — in particular, no specific user dat…

I don't understand how they can say it's unlikely. It's objectively true that they train on de-identified user data (https://openai.com/policies/how-your-data-is-used-to-improve...), and objectively true that they encourage users to submit such data (the setting is on by default). Since it's de-identified I can believe that they can't give a straight yes or no answer here, but it seems more likely than not that at least some amount of his usage became training data. It takes an unusual level of awareness and effort for a user to ensure that all usage is opted out.

Re: Navier-Stokes – Tristan Buckmaster [pdf]

#839

Post by swozey »

27183 wrote: post #818
dgellow wrote: post #371

Earlier quoted context omitted.

If that part of the PDF is true that’s so disgusting, psychopathic behaviour > I said that if OpenAI released its result in the way proposed I would go public with what happened. The reply was, “Why would you ruin your career?” I replied that I am an academic, and asked why he thought going public would ruin my career. The reply was, “If you don’t want me to be nice, then I don’t have to be nice.” Some time later Lev…

This kind of coercive, threatening rhetoric is really not surprising at all. This is how tech companies operate. What stands out to me is the naive lack of operational security on the part of academics, who should know better than to touch this SaaS crap with a ten foot pole.

I'm imagining all of the potential targeted customer lists now. Grab all those sweet .edu, et al logs without anyone thinking the wiser- until you release a stolen solution. What've they got on .gov?

Re: Navier-Stokes – Tristan Buckmaster [pdf]

#840

Post by geocar »

adastra22 wrote: post #473
dgellow wrote: post #375

Earlier quoted context omitted.

What a horrible response from Bubeck. How did he think that would make him and OpenAI look good to tweet that?

Did we read the same tweet? It felt very forthright and level-headed to me. Not at all what I expected.

He seemed to know a lot about what was going on in the state of the art despite not being a fluid dynamics expert or having any in “the project”, and absolutely nothing about what his own employees did with Tristan.

The most uncomfortable piece is where he shows screenshots “proving” his earnestness is unrewarded instead of the chats that are actually being complained about.

That he does so with tremendous gymnastics is perhaps soothing to investors, but every researcher I show this post to today says “see I knew ‘AI’ would steal my research”

So yeah, maybe we are reading different things.

Post reply on HN