Altman says OpenAI will keep publishing math and science research

OpenAI CEO Sam Altman says the company has more mathematics and science research to publish, even as some academics struggle to keep pace with its work.

Speaking on the Cheeky Pint podcast, Altman said humanity deserves fundamental knowledge and should not leave major scientific discoveries unshared. Asked whether OpenAI had unpublished breakthroughs, he replied: “We do. Of course.” He acknowledged that the company’s pursuit of mathematical challenges had “got out of control” at times and said he had empathy for criticism of its methods.

Critics have raised concerns about attribution and verification. The article reports that OpenAI’s work on the exponential time hypothesis (ETH) has not been peer reviewed or independently verified. The hypothesis held that the effort needed to solve a problem increases with its complexity.

The volume of research is also putting pressure on academics. Wired reported that OpenAI recently posted 722 manuscripts addressing 400 problems on GitHub. Thore Husfeldt of the IT University of Copenhagen said the regular announcements can disrupt researchers’ ongoing projects and be psychologically taxing. He also described the findings as exciting, saying they had illuminated an entire area of mathematics.

Researchers quoted by Wired expressed both enthusiasm and concern. Linköping University professor Peter Jonsson said the work can teach researchers a lot, while UC Santa Barbara professor Daniel Lokshtanov called its impact “really cool and terrible at the same time.”