OpenAI posts 700+ AI math preprints, mathematicians revolt

On 6 October OpenAI uploaded more than 700 AI-generated mathematical preprints to GitHub, aimed at hundreds of high-profile open problems, Nature reports. The batch did not include solutions to any of the five remaining Millennium Prize Problems.

The company framed the release as part of its research roadmap. 'The company's goal is to provide tools for mathematicians and scientists,' OpenAI said, adding that it evaluates its models 'against difficult frontier problems in all sciences.'

Reactions among mathematicians split hard. Alvaro Lozano-Robledo, a number theorist at the University of Connecticut, Storrs, called the sheer volume 'staggering' and asked, 'Why is OpenAI trying to solve hundreds of math problems to be released all at once?' Ursula Martin of the University of Oxford described the dump as 'tossing the community a messy first draft and expecting us to clean it up.' Michael Harris of Columbia University went further: 'I will not be reading these papers.'

Not every voice was hostile. Levent Alpöge, a mathematician at Anthropic, told Nature it was 'obviously the most significant moment in mathematical history,' while also warning of 'sad stories' for researchers whose projects had been pre-empted by the OpenAI release. Three researchers we track shared the Nature piece into our feeds the same day it ran.

Nature reports that 'some were incensed at what one mathematician called a slopocalypse, even if the mathematical content could end up being formally correct.' OpenAI's preprint on the Navier-Stokes equation, released in September, solved one of the original seven problems.