News
AI Advances Spark Debate on the Future of Mathematics
AI breakthroughs in mathematics, including proofs of major problems, raise questions about human understanding and the role of researchers.
Advancements in artificial intelligence (AI) are reshaping the field of mathematics, sparking both excitement and existential questions among researchers. The release of hundreds of mathematical manuscripts by OpenAI in October 2026 has intensified discussions about the role of human mathematicians in an era where AI can solve longstanding problems.
AI Solves Decades-Old Problems
One of the manuscripts from OpenAI provided a proof for a problem that Cornell mathematician Justin Moore had worked on for 19 years. While Moore plans to celebrate the resolution with colleagues, he described the experience as feeling "more like a wake than a celebration," according to his colleague Steven Strogatz. Similarly, mathematician Levent Alpöge announced in November 2026 that he believes he has proved the Riemann hypothesis, a Millennium Prize Problem, with the help of AI. Alpöge described the experience as both a loss and a new dawn for science.
Challenges of Understanding AI-Generated Proofs
Dan Boneh, senior research adviser at a16z Frontier and head of Stanford’s applied cryptography group, posed a thought experiment about the implications of AI-generated proofs. He asked whether humans could ever fully understand a billion-page proof, such as one that might resolve the P vs. NP problem. Even with tools like Lean, a programming language for verifying logical steps, the task of making such proofs intelligible would require extensive collaboration among mathematicians and computer scientists.
Reframing the Role of Mathematics
Researchers like Ittai Abraham of a16z Frontier emphasize the dual frontiers of mathematics: discovery and understanding. Abraham referenced a 1994 essay by mathematician William Thurston, which argued that the true progress of mathematics lies in human understanding, not just the quantity of theorems. This perspective is echoed by efforts to simplify AI-generated proofs, such as the work of Sang-hyun Kim and Yash Lodha, who provided a clearer explanation of a result from OpenAI’s manuscripts.
Collaboration Between Humans and AI
Despite concerns, many researchers see AI as a powerful collaborator. Scott Duke Kominers, a Harvard professor and a16z Frontier researcher, shared how AI helped him solve a problem he had been working on since 2007. He noted that the use of AI tools did not diminish the joy of discovery for him or his peers. Similarly, mathematician Frank Vallentin described how an AI model helped him resolve a lattice covering problem he had studied for years, leading to a simplified and elegant proof.
As AI continues to advance, researchers are reimagining the future of mathematics. While machines may accelerate discovery, the human role in understanding, teaching, and applying these insights remains indispensable.