A mathematician at artificial intelligence company Anthropic has claimed to have proved the 167-year-old Riemann hypothesis, one of the most famous unsolved problems in mathematics, using the company's AI model Claude, Al Jazeera reports.
Levent Alpoge, a member of Anthropic's technical staff, announced the claim on Friday in a four-page explanation published on his website, saying he believed he had discovered a proof while working with Claude.
If independently verified, the breakthrough would resolve a mathematical problem that has challenged generations of researchers since German mathematician Bernhard Riemann first proposed the hypothesis in 1859.
However, mathematicians are still examining the claimed proof, and no independent expert has yet reviewed the entire argument.
Alpoge said his work establishes the generalised Riemann hypothesis, a broader version of the original conjecture that applies to a wider family of mathematical functions.
Rather than publishing a conventional research paper, he released the proof in Lean, a computer programming language and proof assistant that allows mathematical arguments to be checked step by step.
Acknowledging Claude's contribution, Alpoge said the AI model had significantly outpaced him towards the final stages of the work.
The scale of the proof, however, presents a major challenge for researchers attempting to understand and verify it.
Kevin Buzzard, a professor of pure mathematics at Imperial College London, said the argument consists of approximately 17 million lines of Lean code, potentially equivalent to tens of thousands of pages of conventional mathematics.
"We don't even yet know how much of this mathematics is new ideas, how much of it is just sort of standard stuff the machine just had to develop," Buzzard told Al Jazeera.
He said mathematicians around the world were working to translate sections of the computer code into conventional mathematical language to understand the underlying argument.
Buzzard expects researchers to gain a clearer understanding within a week or two.
The Riemann hypothesis concerns the distribution of prime numbers, which are numbers divisible only by themselves and one, such as 2, 3, 5 and 7.
Although prime numbers appear irregularly across the number line, the hypothesis suggests that their distribution follows a deeper mathematical pattern.
That pattern is connected to the Riemann zeta function, a mathematical object introduced by Riemann in 1859.
Abhishek Saha, a professor of mathematics at Queen Mary University of London, described the hypothesis as the most famous and important unsolved problem in mathematics.
The problem is among the seven Millennium Prize Problems identified by the Clay Mathematics Institute, each carrying a $1 million award for a correct solution.
A proof would have implications across several branches of mathematics, as hundreds of mathematical results have been established on the assumption that the hypothesis is true, Saha said.
It could also have implications for computer science and internet security.
Buzzard said some algorithms currently used in computing are believed to work correctly based on assumptions linked to the generalised Riemann hypothesis.
He added that cryptographers and mathematicians have long relied on the hypothesis because of the extensive evidence supporting it, although it has never been formally proved.
The claimed breakthrough comes amid a series of advances in mathematical research involving artificial intelligence.
An earlier AI-assisted result demonstrated that at least 67.25 percent of a particular set of mathematical values followed the pattern predicted by the Riemann hypothesis. The latest claim, if verified, would establish the result across the entire set.
Saha said AI was transforming mathematical research by helping researchers tackle problems previously considered beyond reach and potentially enabling the development of new mathematical theories.
He argued, however, that the achievement should primarily be credited to Alpoge rather than Claude alone.
"Others with similar access to frontier AI models and computing resources were not able to come up with a proof of the Riemann hypothesis," Saha said, adding that Alpoge's ability to direct and coordinate the AI model was central to the claimed achievement.
The difficulty of the problem has long been attributed to the limitations of existing mathematical techniques.
Saha said mathematicians had spent more than 160 years developing sophisticated theories in attempts to prove the hypothesis, but none had been sufficient.
He said he had previously expected a solution to remain years away and described the new claim, if correct, as a major challenge to existing expectations about mathematical research.
Despite the potential significance of the development, experts caution that the proof's validity and mathematical significance remain to be established.
Buzzard also rejected suggestions that AI could make human mathematicians redundant, describing the technology as a powerful new tool rather than a replacement for researchers.
Comparing AI with the invention of the telescope, he said the technology was allowing mathematicians to explore problems in ways that were previously difficult or impossible.
He added that AI still depended on humans to identify meaningful questions and direct its work.
As AI systems become more widely available, Buzzard expects their role in mathematical research to expand beyond a small number of technology companies.
For now, the mathematical community is waiting to determine whether Alpoge's claim represents a genuine solution to one of mathematics' longest-standing problems.