AI makes breakthrough in understanding prime number distribution

In a recent development in mathematical research, Anthropic announced that its large language model, Claude, made a notable advancement related to the distribution of prime numbers. This announcement came in a blog post detailing the findings of a paper written by an unreleased version of Claude. The research focused on the Riemann zeta function, a complex equation that has puzzled mathematicians for nearly 170 years. The Riemann hypothesis, which posits that the only way for the output of this function to equal zero is when the input variable's real part is exactly 1/2, remains unsolved despite a $1-million prize for a solution. Claude's approach, however, did not attempt to prove or disprove the Riemann hypothesis directly. Instead, it explored a related question regarding the zeros of the zeta function. Mathematicians have established that at least 41.6 percent of these zeros have a real part of 1/2, but Claude's research increased this lower limit to 67.2 percent. This finding is significant, as it suggests that AI can contribute meaningfully to mathematical research rather than merely solving predefined problems. The process involved Claude testing 650 different ideas, with the assistance of 60 agentic versions of itself to vet the various approaches. This collaborative effort highlights the potential of AI in tackling complex mathematical challenges, as noted by mathematicians like James Maynard from the University of Oxford and Andrew Sutherland from the Massachusetts Institute of Technology. While the Riemann hypothesis remains elusive, Claude's contribution marks a step forward in understanding prime number distribution, showcasing the evolving role of AI in scientific inquiry.