OpenAI made headlines on Tuesday with the announcement that its internal AI model has successfully solved a longstanding problem in mathematics. The model produced a Lean-verified proof that the Navier-Stokes fluid equations can “blow up,” a significant milestone in the field. This achievement was the result of 10,000 coordinating AI agents working over 88 hours to validate the proof.
However, the celebration was marred by accusations from NYU mathematician Tristan Buckmaster, who claimed that OpenAI had acted unethically in the process. Buckmaster alleged that OpenAI’s Sébastien Bubeck had prior knowledge of his and Anthropic researcher Levent Alpöge’s unpublished proof and had pressured them over credit before making the announcement.
According to Buckmaster, he and Alpöge had been working on solving the fluid blow-up problem for nearly a year using AI models. They had arrived at a solution by August 22, which they shared with a mathematician at OpenAI on September 3. Buckmaster claimed that Bubeck then informed him that OpenAI had already produced a 100-page proof using their internal model, which closely resembled the approach taken by Buckmaster and Alpöge.
Buckmaster further alleged that Bubeck offered him two options: either OpenAI publishes their proof the day after Buckmaster’s team does, or Buckmaster writes up the paper alone, excluding Alpöge due to his affiliation with a rival lab. When Buckmaster chose to go public with his findings, Bubeck reportedly questioned his decision and implied that it could harm his career.
Despite Buckmaster’s claims, OpenAI, Bubeck, and CEO Sam Altman have vehemently denied any wrongdoing. Bubeck shared texts that he says demonstrate his offer to coordinate the release of the proofs with Alpöge and maintain transparency throughout the process. Altman defended Bubeck, stating that he had acted with integrity and offered to collaborate once it became clear that Buckmaster’s team had only solved part of the problem.
In a separate statement, OpenAI clarified that they had not seen Buckmaster and Alpöge’s work before publication and had not accessed any specific user data related to their research. The controversy surrounding the announcement highlights the complexities involved in groundbreaking research and the importance of transparency and collaboration in the scientific community. OpenAI, a prominent artificial intelligence research lab, is making waves in the math world with recent developments in AI-assisted mathematical proofs. Fields Medalist Terence Tao praised the work of researchers Buckmaster and Alpöge as a “remarkable achievement.” This comes hot on the heels of another groundbreaking AI-math story from the same team at OpenAI, where they claimed to have produced a computer-checked proof of Fermat’s Last Theorem.
The most significant result from Buckmaster and Alpöge, which closely approaches the $1 million problem, is still awaiting publication pending a final check using the Lean theorem prover. As of now, no external experts have reviewed the extensive 100-page proof.
While OpenAI has not explicitly confirmed the possibility of solving the Navier-Stokes existence and smoothness problem, the recent successes in AI-assisted mathematics have sparked excitement within the scientific community. The intersection of artificial intelligence and mathematics has the potential to revolutionize how complex problems are approached and solved.
The ongoing research at OpenAI highlights the increasing role of AI in advancing mathematical knowledge and pushing the boundaries of what is achievable in the field. As we await further developments and validations of these groundbreaking results, it is clear that the future of mathematics may be shaped by the collaboration between human mathematicians and artificial intelligence systems.
