Mathematician and Anthropic employee Levent Alpoge claims to have discovered a proof of the Riemann hypothesis with the help of artificial intelligence. He worked for around six weeks with a non-public version of the Claude chatbot, but the claim has still not been confirmed by other mathematicians.
“This is the beginning of something much bigger for science,” Alpoge says. In his words, the decision to publish the results was a personal one and was not connected to an attempt to draw attention to Anthropic.
What Alpoge claims
The mathematician says he presented the complete proof in Lean – a system that allows mathematical proofs to be formally verified by a computer. He points out that the material created in the process is more than 250,000 pages long.
However, the size of the proof is one of the reasons why verifying it will be difficult. Mathematician Terence Tao warns that understanding and checking such a vast body of material will take time.
“I expect the process of understanding it to take some time, although in this case at least we have an author we can communicate with and who can help,” Terence Tao commented.Alpoge himself acknowledges that the timing of the announcement is awkward, as it came immediately after Anthropic’s stock market debut. “If I could have, I would have finished all of this much earlier,” he says.
What the Riemann hypothesis is
The hypothesis was formulated in 1859 by German mathematician Bernhard Riemann and is one of the best-known unsolved problems in mathematics. It is related to the Riemann zeta function and the distribution of prime numbers – numbers that are divisible without a remainder only by 1 and themselves.
In more precise terms, the hypothesis states that all non-trivial zeros of the zeta function have a real part equal to 1/2. Proving it would have implications for a number of areas of higher mathematics that rely on assumptions about the distribution of prime numbers.
What has been confirmed so far
At present, there is no independent confirmation that the Riemann hypothesis has been proved. Alpoge’s claim itself is not equivalent to an accepted mathematical result until the proof has been checked by specialists and withstood independent scrutiny.
Anthropic has already reported progress by Claude on a related problem. The company said that the model had raised the known lower bound for the proportion of zeros of the zeta function located on the critical line from 41.6% to 67.2%. In the same announcement, Anthropic clarified that Claude had not proved the Riemann hypothesis itself.
Therefore, for now, the hypothesis officially remains unsolved. If the proof presented by Alpoge is confirmed, it would be a significant result both for mathematics and for the use of artificial intelligence in scientific research.