OpenAI riles mathematicians with Navier-Stokes solution

OpenAI has claimed a solution to the Navier-Stokes problem, a legendary mathematical puzzle that has remained unsolved for roughly 90 years. To crack the fluid dynamics problem, which carries a $1 million Millennium Prize reward, the company deployed an unreleased internal AI model described as more powerful than its new GPT-6 Astra. The system, which began training on August 28, achieved the breakthrough by running alongside 10,000 concurrent agents. This milestone was part of a broader push that yielded solutions to 10 long-standing mathematical problems, demonstrating unparalleled results on advanced benchmarks.

However, the rapid-fire breakthroughs have triggered a severe reputational crisis within the academic community. Researchers accuse OpenAI of aggressive, competitive tactics that violate traditional academic norms. Mathematician Andreas Thom raised concerns that OpenAI unethically used ChatGPT interactions with him and Gabor Kun regarding non-sofic groups to train its models without transparency. Abhishek Saha, a professor at Queen Mary University of London, noted that the company has engaged in behavior that human mathematicians would generally avoid, including rushing to scoop rival researchers.

For mathematical practitioners, this aggressive AI integration threatens to disrupt how research is conducted, credited, and shared. Fields Medalist James Maynard of the University of Oxford expressed deep concern over how the traditionally slow-moving discipline will adapt to a sudden influx of automated discoveries. In response to the growing dread, OpenAI has established a new independent advisory panel of elite mathematicians. This group is tasked with guiding the company on how to present future discoveries and manage its relationship with the wider academic community, though many researchers remain skeptical about how much influence the panel will actually wield.