AI and mathematician both claim solutions to 167-year-old Riemann Hypothesis
Tech Desk
It has been a significant week for Levent Alpöge, a mathematician who works for Anthropic, the first publicly traded frontier AI lab. He appears to have solved the Riemann hypothesis, a problem dating back to 1859 and one of the Millennium Prize math problems.
Advertisement
“The precise timing is horrible,” Alpöge noted in an essay where he explained messaging Anthropic CEO Dario Amodei on Sunday night. He wanted advice on how to announce that he had discovered a Lean-compilable solution to the hypothesis, which suggests a hidden structure exists beneath prime numbers. If the result is valid, it represents a profound advancement in understanding the universe.
Experts have been debating the findings on a Zulip chat, with number theorist Chris Birkbeck from the University of East Anglia commenting on the technical nature of the work. He noted that the cohomology bound proof utilises Galois representations, base change, Jacquet–Shalika, Clozel’s purity, Bernstein–Zelevinsky, the main theorem of CM, and Deligne–Ribet. This post received 20 reactions from readers.
While the community will need time to fully grasp the proof, there is confidence that the formalized Lean result will stand. Lean is an open-source coding language that allows for the formal verification of mathematical proofs. MIT mathematician Andrew Sutherland and other participants in the Zulip discussion have confirmed that Alpöge’s solution compiles successfully.
Alpöge expressed slight guilt about publishing the proof without a clear explanation for humans, yet his lifelong obsession with the problem meant he felt he “could not live” if he delayed the release. He described his process as a “stream-of-consciousness jaunt through the experience of finding all the pieces of the argument through incredible confusion with Claude.” He added that the AI model “flew, and danced, and sparkled near that finish line.”
Advertisement
It remains necessary for humans to understand and explain the result, a task Alpöge has pledged to undertake. Additionally, Alpöge and NYU mathematician Tristan Buckmaster released another proof concerning the Navier-Stokes equations, which contributed to the recent competition between mathematicians and AI labs.
Separately, Stanford professor and American Mathematics Association president Ravi Vakil shared a proof of the General Riemann hypothesis by Stanford postdoc Alessandro Masullo. Vakil stated, “it is all human, with no AI.” He explained that the intention was to bring Masullo to Stanford for a year to refine the document, but they decided to release the draft now.
Because this proof is not written in Lean, it is premature to determine if it is correct or understandable. In his introduction, Masullo mentioned that the work represents 12 years of effort. He wrote, “if any message at all this work is meant to send to the mathematician who feels discouraged by the pace and tone of the current climate, it is this: we are still here. We are very much alive. We are not going anywhere.”
Independent sources have verified that Alpöge’s Lean proof is correct.