OpenAI's Navier-Stokes proof questioned by mathematicians

openai presented a 166-page potential solution to the navier-stokes problem in early september, claiming it followed the approach of spanish mathematicians diego córdoba and luis martínez-zoroa. however, several groups of experts, including mathematician gonzalo cao-labora of epfl, disagree and say the similarities are only surface-level. openai used a formal verification language called lean and deployed about 10,000 model instances for roughly 88 hours and 130 billion tokens to develop the potential solution. openai researcher noam brown says nobody knows what the massive agent swarm contributed to the result. the math community is concerned about the implications of ai-generated solutions and the lack of transparency in openai's methods.