OpenAI's Math Advisory Group Stumbles Out of the Gate — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- AGMAI — the Advisory Group on Mathematics and Artificial Intelligence, a nine-member body of elite mathematicians — was announced September 21 via a guest post on Terence Tao's blog, with OpenAI publicly stressing the group's independence.
- Martin Hairer, an AGMAI member and professor at Imperial College London and EPFL, acknowledged OpenAI's own announcement "didn't really help" establish the group's independence, adding that the company is "obviously very good PR people."
- OpenAI says its unreleased internal model has now resolved "more than 100 long-standing open problems across most areas of mathematics" and is hinting at another Millennium Prize problem nearing solution, fueling dread among researchers.
- Álvaro Lozano-Robledo, a University of Connecticut math professor, said OpenAI's habit of dropping results without engagement "is not how any academic behaves" and argued that "solutions are often less important than the understanding that comes with them."
- Hairer accused OpenAI of altering manuscripts "sneakily" after release in response to criticism, calling the practice "shoddy" and "really bad and sloppy scholarship."
- AGMAI's formation came partly in response to OpenAI's prior Millennium Prize problem announcement, which descended into fights over credit, scooping, and its treatment of mathematicians — and Hairer conceded the group started "two days ago" with "not like we have a big master plan."
Why it matters: OpenAI says it is sitting on more than 100 mathematical results it plans to release, and the company depends on the very mathematicians it has alienated to evaluate and contextualize those breakthroughs. AGMAI — the bridge meant to repair that relationship — has existed for days, has no master plan, and one of its own members is already publicly calling OpenAI's conduct "shoddy," meaning the credibility gap widens before a single new paper lands.
Ask SkimNews
