이미지: The Verge AI
Summary
- James Maynard, an Oxford professor and Fields Medal winner, revealed he has been wrestling with the future of mathematics in the AI era
- OpenAI recently announced it had produced solutions to 10 long-standing mathematical problems that had gone unsolved for decades
- Anthropic also reported partial progress on the Riemann Hypothesis on August 10, using an unreleased research version of Claude
- 인터뷰 대상
- 제임스 메이너드, 옥스퍼드대 교수·필즈상 수상자
- 오픈AI 발표
- 장기 미해결 수학 문제 10개 해법 도출 주장
- 보도 매체
- The Verge, 2026년 8월 11일 발행
- 관련 사례
- 앤트로픽, 8월 10일 미공개 Claude로 리만 가설 부분 성과 보고
A Fields Medalist's Confession
James Maynard is a professor of mathematics at Oxford University and a Fields Medal winner. In an interview with The Verge, he said that over the past year he has "been wrestling with the future of my field." Just days before the interview, OpenAI announced it had produced solutions to 10 long-standing mathematical problems that had stumped academia for decades. According to Maynard, mathematics—traditionally a slow-moving discipline—is now scrambling to adapt in the face of AI.
Can Pattern Recognition Replace Proof?
Just as generative AI learns patterns and correlations from vast datasets to produce text or images, AI systems that solve math problems work by extracting rules from existing papers and proofs and applying them to new problems. The trouble is that mathematics differs from other fields. In science or medicine, proposing a plausible idea has value in itself, but in mathematics, a single broken link in the logic collapses the entire proof. As a result, verifying whether an AI's proposed solution is actually a rigorous proof—or merely plausible-sounding reasoning—has emerged as a new challenge for mathematicians.
This trend isn't limited to OpenAI. Anthropic also announced on August 10 that it had tackled the Riemann Hypothesis using an unreleased research version of Claude. While it did not prove the hypothesis itself, the company claimed it had raised the lower bound on the proportion of zeros of the Riemann zeta function that satisfy the hypothesis, beyond previous results. The sight of the world's leading AI labs racing to deploy their models against the toughest problems in pure mathematics would have seemed unusual just a year or two ago.
So What Changes Now
The fact that a top-tier mathematician like Maynard is publicly airing an identity crisis suggests that AI is moving beyond a mere computational aid and encroaching on the territory of creative proof itself. Whether the solutions to OpenAI's 10 problems will hold up under peer review remains to be seen. What is clear, however, is that mathematicians now face an entirely new task: figuring out how to verify answers produced by AI. A slow-moving discipline has entered a phase where it must redefine itself to keep pace with the speed of the AI era.
