"deidentified data" isn't much to go by. Say I prompted the internal model this way - "Hey there's a solution to a unsolved problem X. The solution uses a less known Method Y so don't bother wasting time with the usual methods. Take papers A, B and C as references. Oh btw, here's the last year's worth of data of all prompt sessions that mention this problem. Pay special attention to the ones that mention Method Y and sub-keywords Z,W".
This is obviously all speculation but the timing is very suspect. If OAI actually did this (and I suspect whatever they did is pretty much close to this), I think it is highly unethical.
This is a bad argument. This is clearly not how it works. And unless Tristan is some truly alien-like savant (and maybe he is), what's necessary to initiate the AI's work already exists in countless published research papers and not exclusively in his head or notes. AIs can survey the sum total of all prior work on a problem and discern reasonable paths for inquiry.
Tristan is acting as if he's working off of an outdated model of AI, similar to primitive chess-playing models that winnowed the search space much more deterministically. If someone this intelligent truly doesn't get that this is not at all what AI is anymore, then maybe there's no hope that we ever understand it.
But I think he does realize this and he's flailing about for counterarguments from a place of bitterness and dejection, accepting even those that are too weak to be defensible. And that is very human and even forgivable.
Even if OAI had zero data from Buckmaster's sessions, this is in very poor taste and highly unethical. You are front running a researcher just to be able to say you did it first? Tao is right - OAI is treating math results like oil. This is the like Exxon getting a whiff of a massive oil field and racing to the punch by deploying their full crew.
I want to be clear that I agree with this view and with Tao more generally. But we're all just yelling at the wind now.
Oh I don't know, maybe something like this?
"Given how seriously this would violate the most fundamental of academic standards, as well as taint the claimed capability behind this result, we take this issue very seriously, and we're launching a probe into identifying whether any of their research artifacts have entered our training set. We have further begun making changes to our UI/UX on all our surfaces, so that it is always clear whether any particular chat, or other user artifact, is eligible for being trained on."
I promise you that if we took their work from ChatGPT and stuck in a bunch of weasel words to give the opposite impression while remaining technically true, I would quit on the spot.
(I work at OpenAI.)
You can also do something like a release of a GPT-OSS v2, where you actually release training data and checkpoints, and do an experiment where you have some held out math problem dataset, then demonstrate how much training it takes on solutions (or partial solutions) to that dataset before the model saturates that test. While of course that would be a test on a much smaller model, it would cost a tiny fraction of the training on your big model, and it could be used to demonstrate just how much effect data contaminaiton like this could have, especially if you did the same experiment on a few different sized of model to show the scaling laws involved.
Pretty much all of math and science history is basically this pattern again and again. I'm sure all of that was rude as well.
It never happens that you wake up one morning and start working on a new problem that came to you in a dream (*unless you are Ramanujan).
This is business as usual for academia, it's amusing to the discussion over it.
From my perspective, this practice is quite bad mannered, unusual, and heavily frowned upon, but I recognize it's possible that these stories might be more common in other fields. Still, I'd appreciate a strong sign that you aren't making these statements up based on secondhand accounts of what 'academia is usually like'.
> this practice is quite bad mannered, unusual, and heavily frowned upon,
Yes, it is.
Maybe you missed my point?
The fact that is "quite bad mannered, unusual, and heavily frowned upon" does not stop it from happening, and it is common for all high profile inventions and discoveries.
This does not disqualify the person doing the scooping, history remembers them as having the credit, and quietly forgets the person who was scooped.
> Pretty much all of math and science history is basically this pattern again and again. I'm sure all of that was rude as well.
was that you were trying to suggest that it wasn't. It seems we actually share similar views here then? I cannot say anything about how common it is, since I've been pretty lucky when collaborating I guess.
If you have made one, and there was never any dishonest competition, you have indeed been very lucky.
I haven't, but the history of science is rather nasty.
Being scooped is a big deal for the one getting scooped.
It has never been a big deal for the one doing the scooping. History is full of math and science results being scooped. For example, we keep calling it Pythagoras' theorem a few thousand years later.
> Our effort began on September 1st after hearing a rumor which we later realized was related to Levent Alpöge, an Anthropic employee, and Tristan Buckmaster, a math professor at NYU. After the completion of our full project and Lean verification (on September 6th), believing from the rumor they also had a solution of Navier–Stokes, we reached out to them to offer a concurrent release of our result and to recognize their priority in a joint announcement. At that point we found out that they had a resolution of the forced Euler problem. In these discussions we offered them visibility into all of the prompts we used and later to see the proof. We recognize the priority of their work on forced Euler and congratulate them on their remarkable mathematical achievement.
Highly persistent agents + vibe-coded security seems like a problem.
This is no different than scooping them.