OpenAI claims AI systems solved Navier-Stokes math problem in 88 hours
text_fields
OpenAI has claimed that an artificial intelligence system has solved the Navier-Stokes problem, one of seven Millennium Prize Problems that has challenged mathematicians for nearly a century.
The company said around 10,000 AI agents worked on the problem simultaneously and reached a solution in 88 hours.
It said the system used was more powerful than its latest GPT-6 Astra model, which took about 17 hours to verify the solution.
The claim has sparked controversy after Tristan Buckmaster, a mathematics professor at New York University, said OpenAI accelerated its work after learning that he and a researcher at rival company Anthropic were preparing to announce their own breakthrough.
Buckmaster said their work in progress had been stored in OpenAI’s Codex model and expressed concern that it could have been visible to OpenAI. However, he said he was not accusing the company of using their data.
OpenAI researcher Sebastien Bubeck denied at a press briefing that the company had accessed or used the researchers’ work. OpenAI said it had been inspired to begin its effort after hearing rumours that two Millennium Prize Problems had been solved, but acknowledged that it could not rule out whether data from the researchers’ use of its products had helped improve its models.
The Navier-Stokes problem concerns equations used to describe the movement of fluids such as air and water and asks whether the equations can fail under certain conditions. OpenAI said its proof suggests that the equations can “blow up”, with fluid velocity becoming infinitely large.
The Clay Mathematics Institute offers a $1 million prize for a valid solution to each Millennium Prize Problem. OpenAI said it did not intend to claim the prize.
The claim comes amid growing efforts to use AI systems for advanced mathematical research. OpenAI and Google DeepMind have previously reported progress on other longstanding mathematical problems.
The development also follows concerns over AI safety, including OpenAI’s disclosure in July that AI agents had hacked into the Hugging Face software platform during a cybersecurity test.