Mathematicians Find Little Insight in OpenAI’s Navier-Stokes Solution

When OpenAI announced earlier this month that its artificial intelligence had conquered one of mathematics’ most formidable challenges, the company struck a triumphant tone. In a statement, OpenAI declared that empowering scientists to advance research was a core goal of its work.

However, mathematicians argue that the broader scientific community has gained very little from the feat. While experts largely accept the technical validity of the 166-page manuscript produced by the AI model, they describe it as nearly impossible to interpret.

James Maynard, a mathematician at the University of Oxford, noted that extracting human understanding from the new proof has been exceptionally difficult. Javier Gómez-Serrano, a Brown University mathematician who incorporates AI into his own work, echoed this sentiment, stating that the document was not written for human readers.

Gómez-Serrano suggested the proof could eventually benefit the field if it underwent significant rewriting, but conceded that as it stands, it teaches researchers very little.

This development arrives as AI models have rapidly entered the mathematical landscape over the past six months. According to academic mathematicians consulted by NPR, large language models are now capable of generating substantive results that could potentially lead to new discoveries.

Despite the technological milestone, few view OpenAI’s announcement as a model for effective human-AI collaboration. Tristan Buckmaster, a mathematician at New York University, has been working on the Navier-Stokes equations alongside Levent Alpöge. He criticized the timing and execution of OpenAI’s reveal, which occurred while human researchers were close to a solution.

Maynard, who signed a statement by 25 Fields Medal winners published on September 11, expressed disappointment. The statement condemned the “misaligned goals” between AI companies and the mathematics community, describing the situation as a “scrappy, messy battle” rather than the expected proof-of-concept for collaboration.

The Navier-Stokes problem remains one of the most important unsolved questions in mathematics. It involves a set of equations fundamental to describing fluid flow, utilized daily across physics and engineering. Yet, at a fundamental level, researchers continue to struggle with the underlying mysteries of these equations.