admin
OpenAI dropped a massive pile of math papers this week — dozens of claimed solutions, hundreds of files on GitHub, and a lot of righteous hubbub from the math establishment. The Association for Human Mathematics (AHM) fired back with a blunt line: “Mathematicians did not ask for this work to be done.” Translation: they don’t like the messenger. Too bad. Taxpayers and progress don’t run on etiquette.
OpenAI’s math dump and what it means
OpenAI published a repository of hundreds of manuscripts and claimed about 90 solved cases from a list of hard problems. Many files include Lean formalizations and compute summaries. Whether every claim pans out is a fair question — independent verification is needed — but the scale of the release matters. A private lab just rushed a big batch of results into the public square. That breaks the old, slow rhythm of peer review and polite academic waiting rooms. It also forces the verification process to happen in public, fast.
AHM’s tantrum and the academic outrage industry
AHM’s statement, reposted by Professor Terence Tao, called the release a “demonstration of power” and urged colleagues to cut ties with OpenAI. Cue the self-important handwringing. If mathematicians want to treat math like a sacred ritual, fine — they can keep their rituals. But most of the people doing math in America work for the public or for universities funded by taxpayers. When a tool promises faster results, whining that someone used it looks less like moral courage and more like gatekeeping.
Why taxpayers get a say
A lot of mathematicians are employed by the federal government or paid with federal grants. That means taxpayers are underwriting the field. If a private company finds a faster route to real answers, taxpayers have a right to ask: Are we getting the best return on our investment? Complaining that “mathematicians did not ask for this work” ignores the point. Science is supposed to deliver results, not theater. If OpenAI’s work holds up when experts check the Lean files and reproduce claims, great. If not, call it out. But don’t pretend process beats progress when federal dollars are at stake.
Verification, transparency, and the right response
Here’s the adult thing to do: verify or falsify. Experts like Professor Tristan Buckmaster and others are right to demand model release or replication before accepting sweeping claims. OpenAI can and should be pushed to share what’s needed for reproducibility. Meanwhile, AHM and academic critics should stop acting like discovery is their private club. If you’re taxpayer-funded, your job is to serve the public interest — not to curate exclusivity. Let the mathematicians vet the work. Let the public decide whether it was worth the fuss. And if OpenAI is wrong, point and laugh. If it’s right, adapt and get to work.
