The mathematics community is grappling with a profound upheaval following the release of hundreds of proofs generated by an advanced artificial intelligence model developed by OpenAI. On October 6, OpenAI published 722 manuscripts—later withdrawing three—that collectively solved 372 mathematical problems long considered open, some remaining unsolved for decades or even centuries. The scale and speed of these breakthroughs have unsettled many professional mathematicians, who now confront questions about the future of their discipline and their roles within it.

Among those reacting strongly was Hugo Duminil-Copin, a French probability theorist and Fields Medal recipient, who said the event had destroyed his field of research and compared it to a mountaineer no longer needing to climb because helicopters can do it instantly. Similarly, Steven Strogatz, a US mathematician, expressed grief over the disruption to the traditional narrative of mathematical discovery, emphasizing the loss of the heroic aspect of being the first to solve a problem.

The landmark moment came when OpenAI’s AI reportedly solved the Navier-Stokes problem, one of the seven Millennium Prize problems established by the Clay Mathematics Institute in 2000, which had remained unsolved for nearly a century. The AI completed the solution in approximately 88 hours, a feat drawing worldwide attention. While the proof’s clarity and style have been subjects of criticism, many acknowledge that concerns about elegance may no longer be relevant in light of the magnitude of advance. The evaluation process for the Navier-Stokes solution is ongoing.

Experts highlight the rapid progression of AI capabilities over recent years. Kevin Buzzard, a professor of pure mathematics at Imperial College, noted that just two years ago, AI systems could not accurately multiply large numbers, yet have since made extraordinary advances. Buzzard himself has invested years trying to develop machine systems that can outperform humans in theorem proving, advocating for transparency from AI developers regarding their results to better assess their claims.

There are differing views on AI’s role in future mathematical research. Marcus du Sautoy, a professor at the University of Oxford, who recently used AI to disprove a decades-old conjecture, sees the technology as a reinvigoration of his work, emphasizing AI’s ability to integrate vast bodies of human knowledge rather than create original ideas. James Maynard, also a Fields Medalist from Oxford, echoes this sentiment but warns that AI may eventually develop independent theories. Despite excitement, Maynard expressed concern over the disruptive impact of AI’s entry into the field, criticizing the influence of commercial interests that prioritize speed and market positioning over the gradual advancement of human understanding.

The head of computer science at Oxford, Leslie Ann Goldberg, described the situation as the most significant crisis in mathematics since the early 20th century’s failure of Hilbert’s program, which attempted to establish a definitive foundation for all mathematics. Goldberg and others have chosen not to rely on AI-generated solutions, instead reaffirming mathematics’ fundamental purpose as the expansion of human understanding rather than mere problem-solving.

The influx of AI-generated proofs prompts difficult questions about the future careers of mathematics postgraduate students, whose traditional paths into research and discovery face uncertainty. The availability of AI tools capable of solving longstanding problems challenges the value and necessity of human effort in this domain. While some believe human creativity will remain essential, the profession must confront how to adapt to these swift changes.

For now, mathematicians worldwide continue to navigate this unsettled terrain, recognizing that the rapid advance of AI in their field heralds a transformation with far-reaching implications for research, education, and employment.