Part of The AI Control Crisis · 11 stories · since Sep 4 · newest 30m ago
OpenAI's math scoops spark mathematician revolt over 'stolen' proofsMathematicians demand proof OpenAI did not use their private work
4 Sep 10 · 14d ago · 11 articles · 7 posts · 21 comments · 6 sources · development 4 of 8
A second mathematician accuses OpenAI of 'dishonesty' about its training data as reporting spreads across outlets questioning where OpenAI's math breakthroughs actually come from.
“since the completion of Navier-Stokes, we have made substantial progress on another Millennium Prize problem”
OpenAI, AI company, via NYT · x ↗Tristan Buckmaster NYU mathematics professorLevent Alpöge Mathematician, Buckmaster's collaboratorAndreas Thom MathematicianOpenAI AI companyJames Robinson Mathematician, University of Warwick
The whole story articlespostscomments the bright band is this development · numbered dots are the others · click one to jump
What was reported 4 claims about this development
-
first by HN Frontpage, 14d ago · also The Verge
-
first by HN Best, 15d ago · also HN Frontpage
-
first by TechCrunch, 13d ago
-
first by Quartz, 14d ago
What people said 23 voices · best of 24 · verbatim
-
This essentially confirms that OpenAI will soon have officially solved the next Millennium Problem. The rumors were true. Via New York Times “In addition, since the completion of Navier-Stokes, we have made substantial progress on another Millennium Prize problem. We are
-
On your second point: there is a more plausible explanation which David Bessis calls the "overhang". The short version is that there is a large amount of relatively low hanging fruits in mathematics, because no human has broad enough knowledge and enough time to try them all. AI is not constraint by that, and therefore can systematically pluck all…
-
An N.Y.U. Mathematician Clashed With OpenAI Over a $1 Million Proof https://www. nytimes.com/2026/09/10/science /tristan-buckmaster-openai-math-navier-stokes.html?utm_source=flipboard&utm_medium=activitypub Posted into Most Emailed @ most-emailed-newyorktimes
-
I think it's a useful analogy to compare OpenAI to a human collaborator. These researchers willingly collaborated with an OpenAI model, giving it ideas, and OpenAI provided useful replies. Then, OpenAI goes ahead and publishes work along the lines of this collaboration, without attributing the researchers. If OpenAI was in fact a human researcher…
-
OpenAI said they sicced this agent army on Navier-Stokes on Sept 1st, while only a couple of days earlier OpenAI's Noam Brown happened to reply to a tweet saying that they had already tried to solve all the Millennium Prize problems and failed... So, it seems either the previous attempt didn't have the training to succeed, or was just not given…
-
A mental model I was thinking about was - I remember when Travis Kalanick was talking about using the chatbot to discuss “vibe physics-ing” on the all-in podcast.And like - I think there’s a presumption you could make that AI models could overfit to asymptote towards just the capabilities and knowledge we currently have.And that would be amazing!…
-
First, OpenAI is not claiming that the model wasn't trained on those sessions. What they've said is “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 data was accessed in order to solve this problem.” and “We did not use their prompts or proofs to…
-
Both things can be true:1. OpenAI when using your chats in pretraining is improving its model’s intuition. The model parameter size is massive, and while the data is OOM larger it is plausible that model remembers stuff about chats that improves its latent representation.2. During RL on verifiable math and massive compute, the model discovers…
-
Even OpenAI's own publication [0] on Navier-Stokes from two days ago appears to contradict "basically solving anything you throw at it". The chart shows a pass rate of ~0.5 (vs. Astra's ~0.2) on "a curated set of open math problems". (Based on the timelines and events described in the publication, I presume that the "Internal Model" in the…
-
Most scientific breakthroughs are simply a continuation of previous work.I feel that these suspicions of mathematicians "seeding" the models' with intuition on how to solve these problems massively overestimates how much their prompts helped the models, and underestimated how much work the models did.Why? We are scared of AI being smarter than us…
-
OpenAI trying their best to put the Navier-Stokes episode behind them by making the GPUs go brrrr. NYT:https://archive.vn/lWzkk> In its Wednesday night statement, OpenAI said: “In addition, since the completion of Navier-Stokes, we have made substantial progress on another Millennium Prize problem. We are working through how to share these results…
-
> Provenance is hard to track - you would hope OpenAI has very good tools for this, but a full data trail of all inputs is difficult to trace through.What would OpenAIs incentive for this be? They've gotten away with scraping everything and getting it ruled fair use. It seems like willful ignorance is an affirmative defense today. Why would they…
-
If the model was trained proper to the conversation with the researcher took place, there'd be no question of tainting the results. But if any amount of training on the model took place afterward, then yes, everything is thrown into doubt (a core problem with considering anything "original" from a model because of how >a % of everything ever…
-
It would be really useful if the researchers disclose their notes and/or chats (or the key pieces thereof) so people can determine how close their work was to whatever the models produced.I mean, now that they’ve been scooped, what value is there in keeping them private? On the other hand, publishing them can bolster their case and help gauge how…
-
How to steal ideas with AI.step 1, identify high value users by net worth, citation count, or number of followersstep 2, select all prompts by high value usersstep 3, invest 10 billion thinking tokens in modeling an objective for each userstep 4, build an RL environment for each userstep 5, rollout 10 billion tokens per environmentstep 6, train on…
-
The leakage wouldn't be from training, but from other uses of Personal Data.As far as I understand it, users can opt out from the training aspect, but they cannot stop their conversations (“User Content”) being used “[t]o improve and develop our Services and conduct research, for example to develop new features”.
-
Andrew Wiles gave 3 lectures, and only at the end of the last one he announced that he solved FLT. Imagine someone from the audience announced in between the second and the third lecture that they proved FLT (using his ideas, obviously).Why is it OK if openAI does it?
-
300 billion tokens is like.. $5-25 million giving range of OpenAI ouput prices, I"m sure they pay less at cost so, I wonder if more $$$ in wage hours have been spend by humans on the problem. My feeling is yes?
-
ClosedAI has every incentive to scoop academics to juice their valuation. Their public statements are worthless, only the incentive 'alignment' matters and theirs will never be on the side of the user.
-
Both can be true:1. OpenAI couldn't have solved the problem without the researchers' private data for training.2. OpenAI models can solve math problems
-
Tangential to the subject, but this is a bluesky post, containing a screenshot of an X post, which itself starts with "in a detailed Mastodon post"...
-
What’s the limit ?Will Microsoft Word publish your novel on Amazon behind your back ?Will VS Code setup a website with your app idea ?
-
I feel that we don’t praise Lean enough. AFAIU it’s what enables LLMs to brute force those problems
All 8 developments of OpenAI's math scoops spark mathematician revolt over… →
Hacker NewsMastodonXNewswiresBlueskyGoogle NewsReddit