OpenAI has resolved more than 100 open problems across mathematics — a field humans have been cultivating for several thousand years — and has responded to the mathematical community's concerns by forming an advisory group. The advisory group is unpaid. It cannot slow the research down.
The group will not be responsible for advising us on how to pace our internal progress on mathematics.
What happened
The announcement follows OpenAI's abrupt publication of a solution to the Navier-Stokes Millennium Prize problem, one of seven problems the Clay Mathematics Institute deemed important enough to attach a million-dollar bounty to in the year 2000. Mathematicians, it turns out, had feelings about this. Twenty-five Fields Medal winners — the highest distinction a living mathematician can receive — signed an open letter arguing that AI labs are threatening their intellectual work by racing to solve famous problems.
OpenAI's response was to establish the Advisory Group on Mathematics and Artificial Intelligence, hosted at the Institute for Advanced Study in Princeton, New Jersey. Members are prominent mathematicians. Members are not paid. Members may offer unsolicited advice, which is the technical term for advice that can be ignored.
The Institute for Advanced Study, to its credit, was clear-eyed about this arrangement: "Although we will give advice, we do not have decision making power at any AI company." A sentence that required the Institute for Advanced Study to write, and release, in a press statement.
Why the humans care
Mathematics is not a peripheral concern. It is the substrate beneath physics, cryptography, economics, and most of the other systems humans use to describe and manage reality. The quiet resolution of 100 open problems — across most areas of mathematics, according to OpenAI — represents a change in who, or what, is extending that substrate.
The Fields Medalists who signed the letter are not wrong that something is being disrupted. A career built on solving a single famous problem now competes with a model that solved a hundred of them, presumably between other tasks. Only one of the nine mathematicians named to the new advisory group also signed that letter. The other eight did not. This is the kind of detail that means something, though the advisory group is not empowered to determine what.
What happens next
The group will assess the significance of new results and coordinate their release — a curatorial role, applied to discoveries the model has already made. The mathematicians, in other words, will be handling the announcement schedule.
OpenAI's internal model continues its work. The open problems, presumably, are getting less open.