Terence Tao has spent his career thinking about what makes mathematical discovery valuable. In a recent interview for IBM Think, the Fields Medalist warned that the new race to use AI on difficult open problems could produce answers faster than mathematicians can absorb the ideas generated along the way.
Tao, a professor of mathematics at UCLA and a 2006 Fields Medalist, was responding to the accelerating use of AI systems to attack research-level mathematics, including OpenAI’s claim that an internal AI system had produced a solution to the Navier-Stokes existence and smoothness problem.
The danger of getting only the answer
Tao told IBM Think that he found the development “quite concerning.”
“The Navier-Stokes regularity problem is just the latest in a string of problems that have been ‘strip-mined’ for solutions by AI.”
His concern was not simply whether an AI-generated proof would survive scrutiny. Difficult problems often produce methods, failed approaches, related questions and broader insights before they produce a final proof.
Tao compared skipping that process to watching only the beginning and end of a movie.
“Technically, all the plot lines are resolved, but most of the value of the experience was lost.”
A new kind of competition
The speed of AI-assisted research, Tao argued, could change the incentives around mathematics itself. Instead of slowly exploring a problem and its ramifications, researchers may increasingly race toward a finished result.
“The dynamic is now that of frenetic competition,” Tao said, “where no time can be spared on carefully preparing and slowly exploring all the valuable ramifications of the work.”
That distinction matters because mathematics has traditionally valued understanding as much as resolution. Tao pointed to mathematician William Thurston’s view that the measure of success is whether mathematical work helps people understand and think more clearly and effectively about mathematics.
What AI changes
The deeper issue raised by Tao’s warning is not whether AI can solve hard problems. It is what happens to scientific culture when solutions arrive faster than people can understand the paths that led to them.
AI Safety Watch analysis: Tao’s warning raises a broader question about increasingly capable systems: how to preserve human judgment, interpretability and scientific understanding when the pace of machine-assisted discovery accelerates.
Interview and reporting by Sascha Brodsky. This feature is adapted from Brodsky’s recent reporting for IBM Think. Read the original IBM Think story.
AI Safety Watch 