This is an important week for Levent Alpöge. The mathematician saw his employer, Anthropic, become the first publicly traded Frontier AI Lab. He also appears to have solved one of the most important problems in mathematics.
“Precise timing is terrible,” he wrote in an essay announcing his discovery, describing how he sent a message to Anthropic CEO Dario Amodei last Sunday night, asking him how he would announce that he had found a lean-compilable solution to the Riemann Hypothesis, one of the Millennium Prize math problems dating back to 1859.
This hypothesis infers that there is a hidden structure underlying the appearance of prime numbers. If the claimed results are true, we are entering a realm of sublime mystery in the universe.
Perhaps the best way to understand how interesting this is to the math community is to see the experts in Zulip chat baffled by the results. Mathematicians like Chris Birkbeck, a number theorist at the University of East Anglia, have posted thoughts like…
The cohomology limit is proven by what looks like an automorphism lifting argument. It purely uses the existence of Galois representations attached to the regular algebraically polarized canine automorphic representation of GL_n over the CM field, along with basis change, Jacquet-Shalika, Clozel purity, Bernstein-Zelevinsky, the main theorem of CM, Deligne-Ribet, etc.
I myself learned math in college calculus, so I can’t help but decipher exactly what it means.
But 20 readers responded to this comment 🤯.
It will take some time before the math community feels they understand the proof, but most of the community seems convinced that the formalized Lean results are valid. Lean is basically an open source language that helps verify mathematical proofs.
In his essay, Alposi says he feels a little guilty about publishing evidence without a truly human explanation, but says he had been obsessed with the issue since childhood and felt he could “not live” if he waited to publish his findings.
Beyond the AI euphoria, Alpöge described an “experiential stream-of-consciousness jaunt through incredible confusion with Claude to find all the pieces of the argument,” and how the model “flew, danced, and shone near the finish line.”
But for all their actions to have any meaning, we must wait until humans understand the consequences and are able to explain them to each other. This is something Alpöge is committed to pursuing.
Alpöge’s collaborator, New York University mathematician Tristan Buckmaster, also published another important proof related to the Navier-Stokes equations. The Navier-Stokes equations are the problems that helped start a race between frontier laboratories and mathematicians this summer to solve open problems.
Meanwhile, Ravi Bakir, a Stanford professor and president of the American Mathematical Society, posted a proof of the general Riemann hypothesis written by Stanford postdoctoral fellow Alessandro Masullo, saying, “This is all human work, no AI. The hope was to take Alessandro to Stanford for a year to refine the text and publish it once it was run with enough people. Now it seems best to publish the current draft.”
In a brief introduction to the proof, which he said was the result of 12 years of research, Masloh wrote, “If there is a message this work sends to mathematicians discouraged by the pace and tone of the current state of affairs, it is this: We are still here. We are very much alive. We are not going anywhere.”
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
