On October 6, OpenAI released a vast collection of more than 350 artificial intelligence–generated mathematical proofs and findings, stirring considerable response across the global mathematics community. The submissions reportedly address a wide array of longstanding open problems across various mathematical disciplines, including algebra, number theory, theoretical computer science, mathematical logic, and topology.
Among the most notable results is a proof of the “quasi-Riemann hypothesis,” a conjecture related to the famous Riemann hypothesis, one of the five Millennium Prize Problems established to celebrate significant challenges in mathematics. While none of the five remaining Millennium Problems were fully resolved, OpenAI’s work appears to make progress in areas tangential to all of them. Ken Ono, a professor at the University of Virginia and founding mathematician at Axiom Math, a company specializing in AI mathematics, suggested that the quasi-Riemann proof could yield a wealth of consequences that may pave the way for future advances, although its direct impact on solving the original hypothesis remains uncertain.
The release prompted a mixture of astonishment, excitement, and apprehension among mathematicians. Alex Kontorovich, chair of the mathematics department at Rutgers University, lauded one AI-generated proof as worthy of a Fields Medal, considered the highest honor in mathematics. Martin Bridson, president of the Clay Mathematics Institute, described the scale and depth of the findings as “breathtaking,” noting that advancing the frontiers of mathematics so dramatically in a single day would have been unimaginable until recently.
At the same time, concerns were raised about the quality, clarity, and verification process of the results. Bryna Kra, a mathematician at Northwestern University and former president of the American Mathematical Society, emphasized the need for thorough review and skepticism. She observed that the volume and complexity of the materials would require extensive time and effort to verify, as traditional peer review mechanisms have effectively been bypassed. Kra also warned that the way the results were released could disrupt the existing academic ecosystem that supports mathematical research, particularly given that some arguments utilized by the AI may rely heavily on unpublished work.
Jean-Pierre Serre, the renowned French mathematician and recipient of both the Fields Medal and Abel Prize, expressed ambivalence about the role of AI in mathematics. While he acknowledges the technology’s usefulness for tasks such as referencing and error-checking, he cautioned that AI might diminish the creative pleasure of discovery, an aspect vital especially for younger mathematicians.
The upheaval caused by these developments is expected to have significant implications for mathematics education and career trajectories, particularly for graduate students. Thomas Carlson, a doctoral student at Montana State University attending a program at the Simons Laufer Mathematical Sciences Institute (SLMath) in Berkeley, voiced concerns about the evolving value and structure of mathematics degrees in light of this technological shift. Similarly, other young researchers highlighted the intangible, narrative elements of mathematical discovery—struggles, insights, and creative leaps—that cannot be fully captured by automated proofs alone.
OpenAI has since updated its public repository of papers, withdrawing three and correcting others where errors were identified. A spokeswoman stated that addressing mistakes and refining the documents is an iterative process, akin to standard academic practice.
Some mathematicians predict that if these claims withstand scrutiny, AI-assisted mathematical research could herald a new era of industrial-scale problem-solving, fundamentally changing how mathematical knowledge is produced and disseminated. However, the overall sentiment among experts is cautious optimism tempered by a recognition of the significant challenges ahead in validation, integration, and preserving the human dimensions of mathematical inquiry.