Tim Gowers calls OpenAI's Erdős unit-distance disproof "a milestone in AI mathematics" — Fields medalist writes the companion paper
Fields medalist Tim Gowers wrote a companion paper to OpenAI's disproof of the 80-year-old Erdős unit-distance conjecture (covered AM cycle), publicly endorsing the result as "a milestone in AI mathematics" and noting that the model independently discovered a novel construction method based on number theory rather than compiling the proof from historical literature.
The Gowers endorsement is the validation that moves the result from "OpenAI press release" to "mathematics community consensus." Gowers is the most public Fields medalist who has engaged seriously with AI-assisted mathematics through the AlphaProof and AlphaGeometry collaborations in 2024-2025. His framing — that the construction was novel rather than retrieved — is the specific claim that the AM-cycle skeptics would have challenged absent independent verification by a credible mathematician.
The proof itself has been published on arXiv (2605.20579v1) and undergone peer review by the global mathematics community. The contribution is a new construction of point sets in the plane with unit-distance pairs exceeding the previously believed optimal upper limit — which means Erdős's 1946 conjecture about the upper bound on unit-distance pairs is wrong by a specific amount. The Gowers companion paper extends the construction technique and demonstrates that the method generalizes beyond the unit-distance case to a broader family of combinatorial-geometry problems.
OpenAI — OpenAI model disproves discrete geometry conjecture → · Phys.org — AI makes major breakthrough in math problem that had stumped experts for decades → · AI Base — Breakthrough in OpenAI Inference Model: Erdős Unit Distance Conjecture →