He was close to a huge math breakthrough. Then he got scooped by AI.
TL;DR
NYU mathematician Tristan Buckmaster spent months with Levent Alpoge developing an unusual approach to the Navier-Stokes problem, one of the $1 million Millennium Prize challenges. Hours after the pair released AI-assisted work on related equations, OpenAI announced that about 10,000 AI agents had solved the problem in 88 hours. Buckmaster questions whether the agents were fed on his work, since OpenAI pursued the same method.
Nauti's Take
If AI agents really cracked a Millennium Prize problem, that is a potential breakthrough capable of compressing years of research into days. The dispute also exposes the limit: without transparent logs of which sources and prompts an agent system saw, authorship is almost impossible to settle.
Researchers should timestamp their progress early, and labs would do well to disclose the data and search trails behind projects like this.
Sources