Press "Enter" to skip to content

Anthropic mathematician claims AI-assisted proof of Riemann hypothesis

Key takeaways:

  • Levent Alpöge says six weeks of work with Claude produced a Riemann hypothesis proof exceeding 250,000 pages.
  • The hypothesis remained listed as unsolved by the Clay Mathematics Institute hours after the announcement.
  • Mathematicians say checking and understanding the proposed proof will take time.

An Anthropic mathematician says he and the company’s Claude AI chatbot have proved the Riemann hypothesis, a problem about prime numbers that has resisted mathematicians since 1859. The claim could mark a major breakthrough, but experts say checking the enormous proof and understanding how it works will take time.

Levent Alpöge, a member of Anthropic’s technical staff, announced the result Friday in a four-page explanation on his website. He said the work took about six weeks and produced more than 250,000 pages of material. According to NPR, the proof consists of about 17 million lines written in Lean, a language used to check mathematical proofs. Alpöge used a version of Claude that is not publicly available, the BBC reported.

“Claude and I have had a great time together with such breadth in recent months, but this time Claude far outpaced me near the finish line,” Alpöge wrote. “No, Claude flew, and danced, and sparkled near that finish line.”

The hypothesis, named for mathematician Bernhard Riemann, concerns the distribution of prime numbers — numbers divisible only by one and themselves. Hours after Alpöge’s announcement, it remained listed as unsolved on the Clay Mathematics Institute’s website, NPR reported. The institute chose it as one of seven Millennium Prize Problems in 2000 and offers $1 million for a solution.

Martin Bridson, the institute’s president, called the announcement one of “historic significance” that holds the promise of changing perceptions of what mathematicians can achieve. “But the journey from today’s announcement to a place of real understanding will be long and difficult,” he said.

Mathematician Terence Tao likewise said reviewing the proof would take time. “I imagine the digestion process will take some time, though in this case at least we have a human author to communicate with to assist with this task,” he said.

Alex Kontorovich, a mathematics professor at Rutgers University, told NPR that a computer’s acceptance of a proof would be encouraging but would not replace his own review. “Does that mean I understand it? Not yet,” he said. “But I have to do the work.” Alpöge himself said he was presenting a complete argument before mastering the mathematics surrounding it.

The announcement came a day after Anthropic’s stock market debut. NPR reported that the company raised a record $81 billion at a $1.5 trillion valuation. Alpöge said he had messaged CEO Dario Amodei on Sunday about whether he could announce the finding. “The precise timing is horrible,” he said, rejecting what he called a cynical interpretation of the timing.

Other AI-assisted mathematical claims also await scrutiny. In September, OpenAI said its tools helped produce a proposed result concerning the Navier-Stokes equations, which describe fluid flow. The BBC reported that the proof had not been confirmed and that OpenAI later claimed solutions or significant progress on 372 more outstanding problems. A group of mathematics scholars warned that AI companies’ promotion of such claims to demonstrate their products was “detrimental to the science of mathematics.”

Sources

Be First to Comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Share via
Copy link
Powered by Social Snap