OpenAI unleashes hundreds more math results upon a field already in shock

OpenAI has released 372 mathematical results produced by an internal large language model, claiming each resolves or significantly advances a major open problem in mathematics or theoretical computer science. The results, posted to GitHub, include a solution to the four-dimensional Kakeya conjecture, improvements to important computer algorithms, and progress on the Riemann hypothesis. Many have been verified in Lean, a proof-validation language. OpenAI says the model generated almost all results from a single prompt given to a single AI agent, a dramatic contrast to its earlier Navier-Stokes solution, which required a 10,000-agent swarm and millions of dollars in computing. Mathematicians are skeptical, citing OpenAI's history of bold claims and limited transparency. The company has not released the model, prompts, or exact compute times, despite recommendations from an independent advisory group it assembled. Some mathematicians argue that releasing the proofs is beneficial, while others, including Terence Tao, criticize the "insane" pace of AI-generated results. OpenAI admits that its own mathematicians do not yet understand many of the results but insists it cannot slow down because these problems are essential tests of AI progress.

(Source：Scientific American)