OpenAI’s Navier-Stokes math breakthrough sparks dispute over who gets credit

OpenAI claimed to secure a remarkable breakthrough by solving century-old Navier-Stokes, one of the biggest math-related puzzles.

On Tuesday, the ChatGPT maker touted that about 10,000 AI agents worked simultaneously on the Millennium Prize Problem and they reached a solution in 88 hours after an effort that cost millions of dollars.

This is the math puzzle that human experts failed to crack for more than decades. OpenAI claimed that this proof suggests that such an equation can be cracked over time as its GPT-6 Astra model took about 17 hours to verify the solution.

"This is a significant milestone for AI research, and its promise for the world is that even more of our hardest questions would become possible to answer," Mark Chen, OpenAI's chief research officer said.

But this breakthrough does not end there. It sparked a debate about who really gets the credit for solving one of the hardest equations.

NYU professor Tristan Buckmaster and Anthropic researcher Levent Alpöge alleged that OpenAI may have built upon their unpublished work without proper credit to solve a major fluid dynamics problem (related to the Navier-Stokes / Euler equations).

Buckmaster stated they shared their drafts inside OpenAI's Codex, discussed their approach with OpenAI researcher Sebastian Bubeck, and subsequently proposed a joint release that excluded Alpöge.

"I was told the model did not look up user data. I asked again, about training, and I did not get an answer,” Buckmaster wrote in his statement.

He also revealed that the OpenAI scientists threatened to ruin his career if he refused to collaborate with them.

However, OpenAI called these allegations “categorically false,” asserting that its researchers did not see the mathematicians’ work before their public release and no specific user data was accessed to solve the problem.

OpenAI researcher Sebastien Bubeck stated that OpenAI started its working on the problem due to viral rumors regarding Anthropic. Furthermore, Sam Altman clarified that the company even used different approaches while solving the problem.

"To be clear, we did not use their prompts or proofs to prompt our models or direct our agents," Bubeck said.

OpenAI also acknowledged broadly that de-identified user data from products like ChatGPT or Codex are used in general model training, meaning they couldn't entirely rule out that usage data helped improve models indirectly.

The Navier-Stokes math problem is one of the seven Millennium Prize Problems designated by the Clay Mathematics Institute in 2000, describing how fluids including liquid and gases move and mathematically expressed.