29/05/2026
OpenAI’s AI model has autonomously disproved a major long-standing mathematical conjecture.
The conjecture in question is Paul Erdős’ Unit Distance Problem (posed in 1946) — one of the central open problems in discrete geometry. It asks: given n points in the plane, what is the maximum number of pairs that can be exactly distance 1 apart?
For nearly 80 years, mathematicians believed the optimal configurations were based on simple grid-like structures, with the number of unit distances growing only slightly faster than linear.
OpenAI’s model has now shown this is not the case. It discovered an infinite family of configurations that produce asymptotically more unit distances — achieving a genuine polynomial improvement over previous bounds.
What makes this breakthrough especially remarkable is the method: the proof draws on sophisticated tools from algebraic number theory, including class field towers and the Golod–Shafarevich theorem — areas far removed from classical plane geometry.
Independent mathematicians have reviewed and confirmed the validity of the result.
This represents one of the strongest examples to date of AI moving beyond assisting researchers to generating original, verifiable contributions in fundamental mathematics.
Full announcement from OpenAI:�https://openai.com/index/model-disproves-discrete-geometry-conjecture/
An OpenAI model solved the 80-year-old unit distance problem, disproving a major conjecture in discrete geometry and marking a milestone in AI-driven mathematics.