Sam Altman to Infuriated Mathematicians: The Math Slop Won’t Stop

Heather Diehl/Getty Images OpenAI Sam Altman says sorry, mathematicians: the onslaught of AI math slop will continue. The ChatGPT maker sparked an existential crisis in the field by releasing hundreds of AI-generated solutions to math problems that haven’t been solved by humans, and sparked further controversy over how it chose to release its solutions and communicate with mathematicians. But this doesn’t seem to bother Altman all that much. In an interview on the Cheeky Pint podcast on Thursday, he acknowledged the disruption that OpenAI was causing to mathematicians, but said that the work must continue — for the good of humankind, he argued. “I think, on net, the world deserves to understand how the universe works,” Altman said the podcast appearance, as spotted by The Verge. “But I do have empathy for the method.” According to the outlet’s reporting, it’s the first time that Altman has directly acknowledged his company’s impact on mathematicians. OpenAI’s explosive foray into math started when it claimed to have found a solution to the famous Navier-Stokes problem by using 10,000 AI agents that worked 88 hours straight. A mathematician accused OpenAI of unfairly piggy backing off his unreleased research and threatening to ruin his career when he said he told them he would go public with his accusations. Prominent names in the field, meanwhile, began to worry about the potential new paradigm that OpenAI had potentially introduced: wealthy AI companies could now jump on someone else’s ongoing research and beat them to the finish line by throwing ungodly amounts of AI computing power at the problem, and then take credit for it; OpenAI had spent around $10 million just to come up with its Navier-Stokes solution. But that was just the beginning. Soon, it released hundreds of more AI-generated mathematical proofs, a move that was blasted by top mathematicians and even a math advisory group that OpenAI had been consulting with. Critics note how none of the papers have been vetted or reviewed, not least of all because the natural language proofs are borderline incomprehensible. It will take years for mathematicians to unpack all the AI-generated proofs OpenAI has dumped on their laps, while OpenAI, in the here and now, gets to use them as purported proof of their technology being revolutionary. On the podcast, Altman admitted that “we did let it get it out of control for one week on math,” but justified the endeavor because of the importance of what they were finding. “I feel unequivocal on this point: Humanity deserves fundamental knowledge, and we should not be sitting on a major fundamental science discovery,” he said. “The world deserves that.” More on AI: Pope Leo Comes Out Swinging Against AI Personhood