OpenAI Says AI Agents Solved a 90-Year-Old Maths Problem

Swan.my.id - Jakarta, OpenAI says it has found a solution to part of a 90-year-old OpenAI maths problem involving the Navier–Stokes equations after directing about 10,000 artificial intelligence agents to work on it for 88 hours.

The ChatGPT-maker described the result as a milestone for AI-assisted mathematics. However, the solution has not yet been independently verified or publicly accepted by the Clay Mathematics Institute, the US organisation behind the Millennium Prize problems.

The Navier–Stokes equations describe how fluids move and sit at the centre of research into turbulence, a phenomenon that remains incompletely understood. Important aspects of the equations have lacked a proof for roughly nine decades.

How OpenAI approached the maths problem

OpenAI said it began training a new internal model at the end of August. The model quickly showed strong mathematical ability, although it remains available only within the company because it is considered significantly more capable than its latest public release.

After hearing rumours on 1 September that two Millennium Prize problems had been resolved, OpenAI said it assigned thousands of AI agents to investigate some of the remaining challenges. By 5 September, the agents had produced a solution related to the Navier–Stokes existence and smoothness problem.

The agents exchanged nearly three million messages during the effort and generated about 130 billion output tokens. OpenAI estimated that a comparable operation would have cost around $10 million, using its own pricing for output from its most advanced models.

What the OpenAI maths problem result claims

OpenAI said its solution addressed two of the four statements required in the Millennium Prize formulation. The prize offers $1 million for a valid solution to one of the designated mathematical challenges.

Despite describing the work as substantial progress, OpenAI said it did not intend to claim the Millennium Prize. The company presented the result as evidence that its AI systems are improving rapidly in advanced mathematics.

That distinction is important because a mathematical result must withstand scrutiny from specialists before it becomes widely accepted. OpenAI’s work has not yet received independent verification or public recognition from the Clay Mathematics Institute.

Mathematician raises questions about the timing

The announcement also prompted controversy. Tristan Buckmaster, a mathematics professor at New York University, said he and Levent Alpöge, a mathematician working at OpenAI rival Anthropic, had been pursuing solutions to the problem using OpenAI’s Codex tool.

Buckmaster said he learned on 3 September that information about their progress had been passed to OpenAI. He argued that the company began its work on the Navier–Stokes equations only after information about the pair’s research had reached it.

He said he had not yet read OpenAI’s complete proof but felt compelled to publicly explain the sequence of events and communications. His statement included excerpts from emails involving OpenAI and questions about the company’s timing and methods.

OpenAI congratulated Buckmaster and Alpöge on their concurrent work. The company said it had not seen their research through any channel before they released it publicly and that it had not accessed user data during its work.

OpenAI added that it could not completely rule out the possibility that de-identified data from product use had helped improve its models, although it considered that unlikely. The company said its proofs differed substantially and that the exact results were also different.

Why independent review matters

The episode highlights both the potential and the limits of AI in mathematical research. AI agents can explore large numbers of possibilities quickly, but speed alone does not establish that a proof is correct.

For now, OpenAI’s claim remains a reported advance rather than an officially accepted Millennium Prize solution. Mathematicians will need to examine the proof carefully before its significance can be determined.</n