Lean and Mathlib communities certify AI-assisted Riemann Hypothesis proof
A joint statement says Levent Alpöge's 17-million-line Lean proof is correct after independent checkers confirmed it. Mathematicians still have to understand the ideas behind it.
Key takeaways
- The Lean and Mathlib communities say they have full confidence that Levent Alpöge's Lean proof of the Riemann Hypothesis is correct.
- The proof is a Lean library of 17 million lines, released on November 20.
- Dozens of people reproduced the build, and several independent proof checkers verified it, the statement says.
- Alpöge, described by TechCrunch as an Anthropic employee and by The Decoder as a Stanford mathematician, said he found the argument while working with Claude.
- Mathematicians still need to understand the proof, and Alpöge has committed to helping them explain it.
The Lean and Mathlib communities say they have "full confidence" that Levent Alpöge's computer-checked proof of the Riemann Hypothesis is correct, in a joint statement that the Mathlib Initiative and the Lean FRO also endorsed. Alpöge, whom TechCrunch describes as an Anthropic employee and The Decoder as a Stanford mathematician, described finding the argument while working with Anthropic's Claude.
Alpöge released the proof on November 20 as a Lean library of 17 million lines of code, according to the statement. Lean is an open-source language for writing mathematics as code so that a computer can check every step. The Riemann Hypothesis, which dates to 1859, makes a prediction about how prime numbers are distributed and is one of the Millennium Prize problems, The Decoder and TechCrunch report.
How was the proof checked?
The statement says four criteria were met. First, the statement of the hypothesis that was proved is the one already in Mathlib, Lean's shared mathematics library, which experts in Lean and number theory had vetted. Second, dozens of people compiled Alpöge's library and reproduced the same raw proof export. Third, two separate comparison tools showed that what Alpöge's library proves matches Mathlib's version of the hypothesis exactly.
Fourth, several proof checkers verified the exported proof, including Lean's official kernel and checkers written by third parties in different programming languages, the statement says. Two of them, con-leche and con-ron, have been formally proven, with set theory as the base, never to accept an invalid proof, which addresses concerns that Lean's own logic could have flaws. The checks ran on an inert copy of the proof, out of reach of any potential malicious code in the source files. A list of the independent checks is kept on the website of MIT mathematician Andrew Sutherland, according to The Decoder.
What is still unknown?
Being formally checked is not the same as being understood. TechCrunch reports it will take a while for mathematicians to get a handle on the proof, though most appear confident the formal Lean result will hold, and Alpöge has committed to helping mathematicians understand and explain the result. The Lean community statement says it looks forward to work across mathematics that will help explain the new ideas.
In an essay, Alpöge described finding the pieces of the argument "through incredible confusion with Claude," TechCrunch reports.
Did it come sooner than expected?
Yes, by forecasters' estimates. Forecasters on Metaculus had expected a solution around May 2035, The Decoder reports, citing Ethan Mollick. Last month Terence Tao predicted the problem would yield to "a combination of humans and AI around 2050 or so," according to The Decoder.
Separately, Stanford professor Ravi Vakil posted a draft proof of the Generalized Riemann Hypothesis by Stanford postdoc Alessandro Masullo that he said is "all human, with no AI," TechCrunch reports. That draft has not been expressed in Lean, and TechCrunch says it is too early to say whether it checks out.
Disclosure: NextFish's AI drafting desk uses Claude, a model made by Anthropic.
Questions readers are asking
Has the Riemann Hypothesis been proved?
The Lean and Mathlib communities have certified that Levent Alpöge's computer-checked proof is correct. Mathematicians have not yet explained the proof's ideas in human-readable form.
What is Lean?
An open-source language for writing mathematics as code so that a computer can check every step of a proof.
What role did AI play?
Alpöge described finding the argument while working with Anthropic's Claude, TechCrunch reports.
Sources
- #riemann-hypothesis
- #lean
- #mathlib
- #ai-math
- #anthropic
- #claude
Get the day's top AI & tech stories on WhatsApp: Follow the NextFish channel
Get stories like this every morning.
Related
AI2 min
Etsy says it will rewrite AI training terms as long-time sellers quit
New terms let Etsy and its partners use seller content to train AI. After long-time sellers left, Etsy said it would not train models to copy their work and promised new wording.
AI2 min
Google opens its Genie 4 world model to developers in the Gemini API
Developers can now generate playable 720p worlds with sound from a prompt or image, at about 10 cents per second of output, with sessions capped at two minutes.
AI2 min
OpenAI details its defenses against adversarial distillation
OpenAI says new safeguards cut attacks that pull hidden reasoning from GPT-6.2 Astra to under 1% success in its own tests, and that it acts against thousands of accounts a week.
AI2 min
Thinking Machines releases Inkling 2 Mini, a looped open-weights model
The 26.4B-parameter model runs its middle layers three times per token and needs about 20GB of memory, letting it run on laptops, the company says.