AI claimed a solution to the 167-year-old Riemann hypothesis — and a human did, too
First reported by TechCrunch ·
AI assistance is now capable of contributing to solving complex, long-standing mathematical challenges.
Mathematician Levent Alpöge, employed by AI company Anthropic, has announced a solution to the 167-year-old Riemann hypothesis, a problem of significant mathematical importance. The solution has been formalized using Lean, an open-source coding language for expressing mathematical ideas, and appears to be compilable, with mathematicians confirming its successful compilation. Alpöge described using Anthropic's AI model, Claude, extensively in his thought process, referring to a "stream-of-consciousness jaunt" with the AI. He also collaborated on a proof related to the Navier-Stokes equations. Independently, Stanford postdoc Alessandro Masullo has also released a proof for the General Riemann hypothesis, which he states is entirely human-generated, following 12 years of work. Masullo's proof has not yet been formalized in Lean, making its verification status uncertain.
The dual announcement of an AI-assisted and a purely human solution to the Riemann hypothesis highlights a rapidly evolving landscape where artificial intelligence is becoming a partner in scientific discovery. While Alpöge's formalization in Lean offers a verifiable, albeit abstract, proof, the community awaits human-intelligible explanations, suggesting AI as a powerful tool for accelerating research rather than replacing human understanding entirely. The validation of Alpöge's work in Lean, even without full human comprehension yet, signals a new paradigm in how complex problems might be tackled and verified in the future.
Masullo's independent, human-only proof, alongside Alpöge's work, underscores that human intuition and persistent effort remain central to mathematical breakthroughs. The juxtaposition suggests a potential bifurcation in how significant mathematical problems are addressed: one path leveraging AI for rapid exploration and formalization, and another focusing on deep, human-driven insight and communication. This dual approach will likely shape the future of mathematical research, with significant implications for academic institutions, funding priorities, and the very definition of discovery.
AI-written summary. May contain errors.