2026-10-30 · contested story

Dartmouth provost resigns over AI use as mathematicians fight OpenAI proof dump

Two related stories dominate this corpus. The first is the resignation of Dartmouth Provost Santiago Schnell, a mathematical biologist who stepped down in late October 2026 after a faculty-led committee found he used AI to 'meaningfully alter his writing' without disclosure, including in a Washington Post op-ed ironically about AI cheating. The second is a broader disruption in mathematics, where OpenAI's bulk release of 722 (later 719) AI-generated proofs — including work on the Navier-Stokes Millennium Problem — triggered a backlash from mathematicians who formed an 'Association for Human Mathematics' and signed statements accusing AI companies of flouting scholarly norms. Both stories orbit the same question: who can be accountable as the author of scientific work when AI does the drafting or the proving?

How each side frames it

left
"This is a good day for everyone that is a teacher, and a good day for the students who are like, 'Oh, right, up is up and down is down,'"
"I want it clear as day. You cannot use generative AI to write," he said. "Not a little bit. Not now. Not ever."
"they wanted to prove they could massacre us"
"The way they took on a life of their own in the media ecosystem is a stark reminder of the steep price people can pay in a culture that has made originality a kind of moral virtue."
"A machine that does in seconds what you spent years learning to do has not merely surpassed your achievement. It has called into question the wisdom of your investment."
center
"the faculty committee found that Schnell used AI to "meaningfully alter his writing" and "was not forthcoming about his AI use.""
"It's either ironic or inevitable — I can't decide which — that the local college most enthusiastic about staying on top of the technologies known as artificial intelligence should be the first to have an A.I. controversy spin out of control."
"there is not human understanding of them at the point of release, and now the work begins."
""Toxic waste" describes a mode of production. It does not describe the truth value of the products."
"The conference will bring together thinkers who are excited about AI's potential as well as those calling into question how the technology is used."
right
"Who can hang a man in the wild, wild west, which is an accurate description of AI in higher ed right now? Especially at Dartmouth, which has so strongly encouraged [AI] use in the name of 'not being left behind'!"
"Santiago Schnell, who had a year-long stint as Dartmouth's number 2 official, resigned from the Ivy League college after a faculty-led committee found that his use of artificial intelligence in his various writings violated ethical standards."
"(Even his resignation letter seems likely to have been generated by AI.)"
"Was he treated with extra scrutiny because of his disposition?"
"Perhaps a better outcome, for Dartmouth and for us all, would have been to use the revelation of Schnell's AI use to start a serious and transparent conversation on our campuses about what we think the lines [about AI use] are"

What each side left out

The right left out — covered by the Scientific American & New Scientist & Monthly Review & BigGo Finance & NewsCord
The right left out — covered by the The Dartmouth & Inside Higher Ed & VTDigger & The Washington Post
The left left out — covered by the National Catholic Register
The left left out — covered by the Scientific American & Business Insider
The center left out — covered by the The Boston Globe
The right left out — covered by the BigGo Finance

What's actually true?

[verified] A faculty-led committee found Schnell used AI to 'meaningfully alter his writing' in violation of at least one publication's disclosure requirement and was not forthcoming about it.
[verified] Semafor flagged Schnell's August Washington Post op-ed on AI cheating as 100% AI-generated by the detector Pangram.
[verified] The committee did NOT use Pangram or other AI-detection tools; it relied on Dartmouth's enterprise Claude logs and Schnell's emails.
[verified] OpenAI released roughly 722 AI-generated math manuscripts, then withdrew three over a sign error, reducing the total to 719.
[verified] 25 Fields Medalists signed a September 11, 2026 statement titled 'A Severe Misalignment of AI in Mathematics,' later gathering thousands of signatories.
[verified] Terence Tao described mass-produced, uncurated AI proofs as 'toxic waste,' a critique aimed at the mode of production rather than AI's mathematical ability.
[contested] OpenAI announced in September 2026 that its model solved a major aspect of the Navier-Stokes Millennium Problem, amid accusations it may have used NYU/Anthropic researchers' chat logs.
[verified] Dartmouth claims to be the birthplace of AI, where the term was coined at a 1956 summer workshop, and hosted an 'AI at 70: What Must Remain Human?' conference the same week as the resignation.

The narrative clash

Whether Schnell's resignation was a fair outcome or an unfair rush to punish amid undefined norms
Left: This is a good day for everyone that is a teacher, and a good day for the students who are like, 'Oh, right, up is up and down is down,'
Right: I am inclined to the view that there is a lot more for us to figure out before we begin to discipline and punish.
Whether mathematicians have accepted AI or remain in crisis
Left: NYU Math professor Tristan Buckmaster said OpenAI's solutions dump wiped out entire research programs.
Right: The day one reaction was like, 'We're cooked, and our field is over.' ... We've now seen a lot of people say, actually, I'm going to have a different but hopefully better kind of career as a mathematician.
Whether Schnell meaningfully used AI or merely refined language
Left: the faculty committee found that Schnell used AI to "meaningfully alter his writing" and "was not forthcoming about his AI use."
Right: He also added that AI detectors are "known to be unreliable." "There's evidence that they're biased against nonnative English speakers, such as myself," Schnell wrote.
Whether there was a clear rule to break
Left: The line on AI-assisted writing is and always was obvious, and this outcome restates it for everyone on campus.
Right: No one at Dartmouth had drawn that line — the college was loudly encouraging AI use, so punishing one administrator for crossing an unwritten boundary is arbitrary.
The right remedy — accountability or institutional conversation
Left: A resignation was the correct and clarifying result; it restores a standard students and faculty can see.
Right: Resignation wasted the moment; the disclosure should have opened a transparent campus debate about where the lines actually are.

94 sources analyzed

Salon Substack 36Kr Reason Fox News Salon PBS NewsHour Nature Reason City Journal Vox Business Insider Africa Dartmouth Investor's Business Daily The Federalist Salon The New York Times NewsCord Reason City Journal The Boston Globe The Washington Post Monthly Review Investor's Business Daily City Journal Vox The New York Times BigGo Finance The Center Square National Catholic Register The Arkansas Democrat-Gazette San Diego Union-Tribune Phys.org Investor's Business Daily The Federalist The Daily Beast PBS NewsHour India Today Reason The Federalist Vox Scientific American The Critic Investor's Business Daily City Journal The New Republic Business Insider Pioneer Press The Federalist Vox CNN New Scientist The New Republic Los Angeles Times InDepthNH.org Salon U.S. News & World Report FlowingData The Boston Globe The Washington Post The Chronicle of Higher Education The Daily Beast Los Angeles Times Livingston Enterprise The Boston Globe Inside Higher Ed The Dartmouth The Daily Beast CNN FOX 13 Tampa Bay The Daily Beast PBS NewsHour Livemint The Boston Globe Los Angeles Times University Herald The New Republic The Washington Post Concord Monitor The New Republic Los Angeles Times Vocal The Washington Post The Killeen Daily Herald The Washington Post Toronto Star CNN Yahoo CNN Yahoo Tech WCAX News-Press NOW VTDigger Observer Online

Perspectivity reads the coverage — not just who published it.

Read the full interactive analysis →