OpenAI Releases 722 Mathematical Manuscripts; Advisory Group States This Does Not Constitute Recognition
OpenAI一夜甩出722篇数学论文!黎曼霍奇BSD全上阵,数学家:读不过来
OpenAI has made 722 mathematical manuscripts public, organized into 372 groups. These manuscripts are said to have been produced by unpublished internal models using a fixed process, with the papers, source code, and some Lean proofs uploaded to GitHub. The manuscripts cover topics such as the Riemann Hypothesis, Hodge Conjecture, BSD Conjecture, and the Unique Game Conjecture. The advisory group includes three Fields Medalists, but they have stated that this does not imply recognition of the results or the production process.
The article provides details about the number of manuscripts, the range of conjectures they address, and the advisory group's stance, offering readers an insight into the pressure for verification within the mathematics community and the nature of AI-driven research output.
Source: 量子位 · qbitai.com