How AI-generated proofs should reach mathematicians
Scientific American reported that OpenAI’s release ignored the recommendations of the advisory group it had formed. The model, its prompts and its reasoning traces remain private.
Source published · In the Weekly Review of 1 November 2026
Filed under Mathematics, AI policy
The Advisory Group on Mathematics and Artificial Intelligence, which OpenAI organised after criticism from mathematicians, issued a clarifying statement a day before the release. According to Scientific American, OpenAI’s handling of two of the three results went against it, by naming specific experts to shepherd the work into public view, and a paper on the existence part of the Yang-Mills problem has still not been published. The group has called on OpenAI to release it, saying there is no scientific reason for further delay.
Those who accept the complexity result, Lance Fortnow among them, point to the Lean proofs and to the simpler human-written proofs that followed within days. The disagreement is less about whether the results are correct than about transparency and about who chooses the verifiers.