Altman stands by OpenAI math push on Cheeky Pint

Altman's remarks on the Cheeky Pint podcast mark one of the first times he has directly addressed OpenAI's impact on mathematicians, following months of AI-generated results that researchers have described as chaotic and poorly communicated. While he conceded the company needs to improve communication, he drew a firm line: “Humanity deserves fundamental knowledge, and we should not be sitting on a major fundamental science discovery.” That position sits uneasily with complaints this week that OpenAI is itself gatekeeping research. Mathematicians and scientists told The Verge they are frustrated that details for one of three claimed breakthroughs — involving Yang-Mills theory — have not been released, a decision OpenAI attributes to a hand-picked group of experts. The episode also offered a rare look at resources behind the push: Altman estimated spending in the order of hundreds of millions of dollars on math problems, including around $10 million on a Navier-Stokes solution. That scale, he suggested, is deliberate — an internal Slack channel called “Model Name-Hold My Beer” was used to throw the hardest problems at models to probe their limits. Researchers' worry that they may be priced out of their own fields appears unlikely to ease.