I'm seeing a lot of this and it makes no sense, the internal model solved these but give one of the papers to Astra and Opus and I'm sure it will have no problem recreating it.
I don't think they mean this from a "verify this paper" perspective.
How valuable it would actually be to share the model with other mathematicians vs just have OpenAI's mathematicians churn out and clean up results isn't very clear to me though as they don't say how much effort it's requiring from their team to prompt and clean these up vs how much it's bound by "time to run the model" or similar.
> Following an investigation, we have confirmed that Buckmaster’s Codex prompts over the two months preceding this announcement and paper on September 8, 2026, could not have influenced the system in any way, including through training. The OpenAI internal model used for this result was developed through large-scale reinforcement learning on top of a previously pretrained model. Our proofs also differ significantly. In the Euler case, Alpöge and Buckmaster proved a result with external forcing, while OpenAI’s system proved a result without external forcing.
At this point it takes some weird epistemics to think they'd need to outright fabricate the outcome of an investigation when countless other mathematical breakthroughs have been solved the same.
They are certainly biased, but I don't think such bias is strong enough to cause them to lie about facts. We may agree to disagree here.
If by now you still think it's all just hype, it's safe to say you've succumbed to a mind virus that renders you unable to think critically about AI. Otherwise you'd have some level of awareness of just how far this technology has developed, and you should find these developments more than plausible.
A.I. is an extremely broad term. I'm not convinced that the capabilities of these LLMs are what they claim them to be.
That's not to say that advances in machine intelligence can't lead to something that's truly useful or even groundbreaking in the future. I'm just saying that the current technology isn't that and I therefore call it a hype.
This comment makes no sense. You realize how much lsp is used in other editors like vim/neovim? The existence of lsp is exactly what lets custom editors to flourish. Any language/lsp client can add their own extensions too. Any new editor can come up with it's own way of doing things and implement it on top of lsp, as long as the lsp client supports those extra features then there's no problem. I don't know how you can get it so wrong.
> I suspect that the average user doesn’t make a distinction between Uber and driver, and possibly assume that drivers are trained to act in a safe manner.
I feel like the exact same line could be used to argue the opposite point against formal verification.
I'm not saying that proof is inherently bad. If it was free, then I agree it would be good, but my point is that it's not free. Proofs are expensive to produce, maintain, they lock-down flawed implementations, focus on correctness but disregard more important aspects like modularity (I.e. loose coupling, high cohesion). Also; formal proofs discourage change and they create false confidence about reliability because sometimes the bug is in the spec itself, especially as the spec gets more complicated.
I think modularity is a more useful property to aim for in terms of achieving the right degree of correctness over the life of the software, in a practical sense.
Formal proofs can work against modularity if the proof must be rewritten in order to achieve modularity as requirements change over time; which is the reality for most software.
I work on firmware and deep embedded systems. Trust me when I say LLMs are getting just as capable in this domain. Eg LLMs can do RF filter design. Tell the LLM your target metrics and it will give very reasonable design constraints, especially when you come back with simulation results and relevant feedback.
Right, but that is self contained domain. Hook it up to anything and an impedance mismatch will offset the intent of your poles. Failure modes, cross domains, integration, loads, design intersections; AI can build components, but its the sprouting of these components that exponentially magnifies the number of failure modes.
Engineering needs to be failure free, or at least failure preventive via redundancy, and the latter requires the same full systems understands as the former
If they are all wrong, that's when it would backfire.
reply