Researchers are getting tired of OpenAI gatekeeping its breakthroughs

OpenAI announced major breakthroughs in mathematics, computer science, and physics, but the company's approach to sharing these results has drawn sharp criticism from researchers. The mathematics result claims to resolve the Kakeya conjecture in all dimensions; the computer science finding claims to disprove the exponential time hypothesis; and the physics result explores quantum chromodynamics and Yang-Mills theory. Each result has a different release strategy: OpenAI released the mathematics output publicly, planned an external group for the computer science finding, and deferred entirely to outside experts for physics without any public release. Researchers acknowledged the significance of the findings, describing them as "huge" and "major breakthroughs," but criticized OpenAI for handpicking which experts could first examine the work. Constantin Kogler of the Institute for Advanced Study said it "doesn't make any sense" to restrict access, while University of Texas physics professor Andreas Karch called the handpicked group "a disaster." The Advisory Group on Mathematics and Artificial Intelligence (AGMAI), which OpenAI had previously pledged to consult, had warned against exactly this practice the day before the announcement, recommending instead that AI companies release findings openly on scholarly repositories. Georgia Tech professor Xiaoyu He raised deeper concerns, arguing that OpenAI is making a show of responsible release practices while fundamental issues remain: results are being solved without human contribution, and the papers are difficult to read and verify. Stanford mathematician Jared Duker Lichtman acknowledged that no release approach will satisfy everyone.

(Source：The Verge)