In 1818, Daniel Webster declared to the Supreme Court: “It is, Sir, as I have said, a small college, and yet there are those who love it.” Today, Dartmouth’s president is erasing Webster’s Dartmouth in The Atlantic. In “Why I Want More AI at Dartmouth,” College President Sian Leah Beilock writes that universities that don’t embrace AI “guarantee their own irrelevance.” She claims her essay is pedagogical. It isn’t — it’s a marketing strategy for growth.
By November 2024, Beilock called for 1,000 more beds, reinstated standardized testing, mass arrested peaceful protesters and reorganized undergraduate education under a new School of Arts and Sciences.
After Trump’s election, she hired the Republican National Committee’s former attorney, adopted an institutional restraint policy barring public statements, changed Dartmouth’s diversity, equity and inclusion website and organizational chart, cut the libraries’ budget, signed an AI partnership with Anthropic and Amazon Web Services and began expanding the student body — while her spokesperson insists we’ll remain “the smallest Ivy.”
AI is the story that lets Beilock’s blueprint for growth sound like a distinction that sets us apart from the other Ivies as we begin to more closely resemble them.
“AI at 70,” the Dartmouth conference marking the anniversary of the 1956 Dartmouth Summer Research Project that coined the term “artificial intelligence” is billed as an inquiry into “what must remain human.” The lineup reads like a trade show for industries whose claims need scrutiny: Amazon keynotes, Anthropic hackathons — the companies selling Dartmouth its AI infrastructure at an as-yet unreported price — and a “Presidential Session” moderated by Beilock to boot.
What’s missing is as revealing as what’s included. Nita Farahany ’98, one of the world’s leading AI and neurotechnology ethicists, isn’t on the program — even though she was quoted in CalMatters last month warning that neural-monitoring tools could let employers detect when workers are thinking about organizing a union. That omission may be innocent, but her work goes directly to the conference’s stated theme — and an expert on these tools’ potential harms would complicate a stage built around Beilock’s enthusiasm for AI.
Those sounding alarms about what their former companies built were also kept off the program.
On Feb. 9, Mrinank Sharma resigned as head of Anthropic’s Safeguards Research team, writing on X and Substack that “the world is in peril,” citing AI and bioweapons among a “series of interconnected crises.” As Beilock’s essay made the rounds, Jacob Coxon, who spent three years on pretraining research at OpenAI and Anthropic, resigned publicly, writing on X that the two companies “are racing straight to self-improving superintelligence and gambling with our lives” and that the people building it “earnestly believe AI could kill all humans.” Two current Anthropic staffers, including its own alignment science lead, publicly agreed. On Sept. 11, Anthropic chief executive officer Dario Amodei said AI development needs to slow down and be closely monitored.
If people inside Anthropic are worried about where their own work is headed, Dartmouth’s president is complicit in handing a spokesperson for a company chasing a $1 trillion valuation a stage and calling it “academic inquiry.”
Beilock’s claim that AI will “future-proof” Dartmouth graduates collapses the moment one looks at who the future has already stopped working for. She opens by citing Stanford University research showing employment for young workers in AI-exposed fields has fallen 19%, then turns to her own research on math anxiety, suggesting anxiety can be reduced through practice and exposure.
That analogy is elegant — and a category error.
Math anxiety is a psychological state, softened by preparation and repeated exposure. The AI economy isn’t a test a student can practice for until the fear fades — it is a series of decisions employers, investors and tech companies are making about which jobs to automate, which workers to keep and which labor to make cheaper, including roles requiring advanced degrees. Adjunct faculty don’t get tenure no matter how many years they teach. Staff attorneys don’t make partner no matter how many hours they bill.
McKinsey’s talent research estimates that roughly five percent of roles generate 95% of a company’s value — another way of saying the rest are first to go once AI makes cutting easier. A new Dartmouth graduate won’t lose a job for failing to get comfortable with bots and AI agents — they’ll lose it, or never get it, because decision-makers chose to pay for AI tokens instead of salary and benefits. AI fluency was never the variable.
When a structural problem is redescribed as an individual anxiety problem, the burden shifts from institutions to students. If the labor market grows more precarious, the failure is no longer that companies automated entry-level work or that policy failed to distribute the gains — it’s that graduates weren’t “AI fluent” enough. Beilock’s future-proofing promise is less a shield than a sales pitch to kids choosing between offers that look identical in prestige and financial aid.
And there’s an elephant in the room: AI use is measurably eroding the skills it claims to protect.
An August 2026 Common Sense Media survey found 70% of teens now use AI for schoolwork, and nearly 40% say they generate fewer original ideas and feel like they’re missing out on learning as a result. In medicine, researchers have documented “automation bias” among radiologists working alongside AI — a measurable drop in diagnostic accuracy once an algorithm has already flagged a scan as normal. Geoffrey Hinton’s famous 2016 prediction that “we should stop training radiologists” turned out wrong — the field has grown, not shrunk, precisely because judgment and pattern recognition built over years of unassisted practice remain irreplaceable. Those are the muscles automation-bias research says atrophy when practitioners lean on AI too early and too often.
“Practice with AI,” Beilock’s prescription for student anxiety, is the same mechanism researchers say produces “cognitive surrender” in classrooms and diagnostic drift in hospitals.
AI is not inevitable or unstoppable — “a world that promises to be defined by AI” is the propaganda written in Beilock’s essay and baked into “AI at 70.”
Researchers and safety advocates across the spectrum have called for a U.S.-China AI nonproliferation framework, since an unregulated race benefits only the companies racing. Neither OpenAI nor Anthropic have taken steps toward this solution, but even Palantir has called for it.
Scholars like Ruha Benjamin and Carissa Véliz, and journalists like Karen Hao, Tressie McMillan Cottom and Gil Duran, author of “The Nerd Reich,” have long argued that AI’s “inevitability” is a manufactured narrative meant to force public submission, one that holds only as long as democratic societies decline to redirect it through public policy — toward distributing gains broadly, protecting workers and keeping decision-making power out of the hands of companies whose own employees are now resigning in protest.
AI causes environmental damage through ravenous energy and water use. AI data centers are on pace to produce 44 million metric tons of fossil fuel emissions by 2030. Going all in on AI means both abandoning Dartmouth’s commitment to a 60% reduction in greenhouse gas emissions by 2030 and 100% by 2050.
Legal settlements have been reached to compensate Dartmouth faculty for having their research and writings accessed without authorization and plagiarized by Anthropic. Claude is an embodied violation of Dartmouth’s Academic Honor Principle. Yet, Dartmouth is rolling it out as fast as it can.
None of this appears in “AI at 70” or Beilock’s essay — no accident, since it’s harder to sell AI as the ticket to a secure future if the people closest to the technology are saying that no such ticket exists.
Beilock didn’t write an op-ed about classroom AI bans. Her essay is designed to sell a pro-AI Dartmouth to prospective students and donors, and propagandize alumni and the public, just as she moves us away from the competitive advantage of being small: a network in which anyone can reach everyone without gatekeeping.
If we helped students use their years in Hanover to build a true community instead of chasing a feed, then they would have a shot in a world flooded with AI slop. But alas, AI is the next institutional transformation under Beilock’s tenure — the next chapter in her campaign to reinvent Dartmouth.
Webster won his case by appealing to what made the College worth defending. Beilock posits her bigger, AI saturated Dartmouth will still be our College. It won’t be.
We’ll no longer be the “enduring institution” Booz Allen Hamilton once recognized.
We will be a small college no more.
Unai Montes-Irueste is a member of the Dartmouth College Class of 1998, Dartmouth College Alumni Council, and president of the Dartmouth Association of Latino/a/x/e and Caribbean Alumni. He lives in Southern California with his wife and three children. Guest columns represent the views of their author(s), which are not necessarily those of The Dartmouth.
