AI has claimed a solution to the 167-year-old Riemann hypothesis – and a human has done it too

It’s a big week for Levent Alpöge: the mathematician saw his employer, Anthropic, become the first publicly traded frontier AI laboratory. It also seems to have solved one of the most important mathematical problems.

“The precise timing is horrible,” he wrote in an essay announcing his findings and describing how he messaged Anthropic CEO Dario Amodei last Sunday evening asking how to announce that he had found a Lean compilable solution to the Riemann hypothesis, which dates back to 1859 and is one of the Millennium Prize math problems.

The hypothesis assumes that there is a hidden structure underlying the appearance of prime numbers. If the claimed result is true, we enter the territory of the sublime mysteries of the universe.

Perhaps the best way to understand how exciting this is for the math community is to watch experts quiz over the result in a discussion on Zulip. Mathematicians like Chris Birkbeck, a number theorist at the University of East Anglia, publish thoughts like this…

The cohomology limit is proved by what looks like an automorphy lifting argument. It truly uses the existence of Galois representations attached to regular and polarized algebraic cuspidal automorphic representations of GL_n on CM fields, as well as base change, Jacquet–Shalika, Clozel purity, Bernstein–Zelevinsky, CM’s main theorem, Deligne–Ribet and more.

I, for one, delved into mathematics during college calculus, so I can’t help but decode exactly what that means.

But this comment was followed by 20 readers who reacted 🤯.

It will be some time before the mathematics community feels they have a handle on the proof, but most seem confident that the formalized Lean result will hold. Lean is an open source coding language used to express mathematical ideas in code, allowing proofs to be formally verified. Several participants in the Zulip chat, including MIT mathematician Andrew Sutherland, claim to have successfully confirmed that Alpöge’s solution can be successfully compiled.

Alpöge seemed in his essay to feel a little guilty for disclosing the evidence without a truly human-intelligible explanation, but his obsession with the problem since childhood made him feel that he “could not live” if he waited to publish the discovery.

Alpöge described a “conscious escapade through the experience of finding all the elements of the argument through incredible confusion with Claude” with more than a hint of AI euphoria, describing how the model “flew, danced and shimmered near that finish line.”

However, for all this action to make sense, we will have to wait until humans are able to understand the result and explain it to each other, something Alpöge is committed to pursuing.

Alpöge and a collaborator, mathematician Tristan Buckmaster of NYU, also published another important proof related to the Navier-Stokes equations, problems that helped start the race between frontier labs and mathematicians to solve the outstanding problems this summer.

Meanwhile, Ravi Vakil, a Stanford professor and president of the American Mathematics Association, published a proof of the general Riemann hypothesis written by Alessandro Masullo, a postdoctoral fellow at Stanford, noting that “it’s all human, no AI. The hope was to bring Alessandro to Stanford for a year to polish the writing, then announce it after running it by enough people. It now seems best to publish the current project.”

Without being expressed in Lean, it is too early to say whether the proof is verified, let alone whether anyone understands it.

In a brief introduction to his proof, which he said was the result of 12 years of work, Masullo wrote that “if this work is meant to send a message to the mathematician who feels discouraged by the pace and tone of the current climate, it is this: We are still here. We are alive and well. We are not going anywhere.”

This story has been updated to include the fact that Alpöge’s Lean Proof has been independently verified.

When you purchase through links in our articles, we may earn a small commission. This does not affect our editorial independence.

Gn bussni

Logo