OpenAI破解1946年数学猜想——数学家担忧被淘汰
openai推翻了單位距離猜想，其新astr模型解決了十個frontiermath問題，這標誌著ai已達到博士級能力，並對傳統數學家的地位構成威脅。
openai recently disproved the unit distance conjecture from 1946 using ai, and shortly after, human researchers adapted the technique to solve another conjecture. epoch ai announced a second solution to its frontiermath benchmark, while openai's astra model demonstrated the ability to solve ten complex unsolved problems. experts like abhishek saha note that frontier models perform as well as phd students on routine tasks, turning mathematicians into 'conductors.' while some view this as a golden age of productivity, tim gowers predicts the rise of ai theorem provers will completely revolutionize and potentially displace pure mathematical research.