Dartmouth College opened its “AI at 70” conference on Oct. 29 with a morning session titled “The Future Worth Building is Human,” and both speakers repeatedly returned to one capacity they said should remain human: judgment.
Before the main conversation, Dartmouth President Sian Leah Beilock addressed a news item that shadowed the campus that week: Provost Santiago Schnell resigned after a faculty-led review found he used AI in his writing without disclosing it. Beilock said the episode underscored the kinds of issues the conference aims to explore and argued that leading on AI requires “first and foremost having leaders who are modeling the transparency and human judgment we expect from our students and our scholars.” She added the college is “still figuring this out, because there’s no playbook yet,” and would rather set its own norms than leave that work to industry or government.
Judgment and leadership in the AI era
Beilock framed judgment as a central skill for an era in which “anyone can get a convincing answer from AI in seconds.” “The hard skill now is judgment,” she said, and described AI literacy as teaching students “how to know when to pick up the technology and when to put it down.” She said some classes will use AI, some won’t, and some will be about AI; a history professor might require blue-book exams, she said, and “all three are totally valid.” Beilock also said the goal is for every student to “build experience with these technologies.”
She closed with a personal anecdote about learning to code in Lisp as an undergraduate and spending a night searching for a missing parenthesis, an experience she said taught her problem-solving and confidence: “I thought that was the pivotal moment in my undergrad career where I learned I could do something I didn’t think I could do.”
Industry control, tacit knowledge and the limits of guardrails
Beilock introduced Mira Murati, a 2012 Thayer School of Engineering graduate and a former OpenAI executive who now runs Thinking Machines Lab. Murati described her company’s work: building models that customers can continue training on their own data, including two open-weight models called Inkling that it released this summer. The idea, she said, is to let people “truly own the intelligence versus renting it.”
Murati warned about risks tied to how AI is built and who controls it. She singled out the danger of a small number of companies controlling the most powerful systems and cited a less-discussed risk: “loss of intellectual diversity.” “Everyone sort of ends up inheriting the same strengths and the same blind spots,” she said, adding that rather than a world of many competing ideas you can end up with “a convergence of thinking.”
On technical risks, Murati said OpenAI had studied hallucinations and misuse before launching ChatGPT but “didn’t learn how to build stronger guardrails until real users ran into them.” She also said, “we haven’t solved the alignment problem,” referring to the challenge of ensuring AI systems do what people intend.
Murati touched on the conference’s central question but did not provide a definitive answer. She said society will need a view on “what does it mean to have human worth in this time of machines being able to supersede us in some domains,” and called for that conversation to enter the public consciousness. She described much human knowledge as tacit—what people know “sits in their minds,” not “perfectly articulated”—and said current AI systems are missing that. Her vision is for AI that better picks up tacit knowledge and “come[s] towards us and understand[s] us.”
Addressing students, Murati urged depth over speculation about job safety: “Don’t spend college trying to guess which jobs AI won’t take,” she said. “Really put something and go deep into it and really understand it deeply.” She also encouraged students to use available AI tools: “This is an incredibly precious time to have these superpowers in your hands,” adding, “I am jealous.” When asked which human capacities are most important to protect, Murati began with judgment, saying it “doesn’t work that way” to simply hold on to judgment while machines do everything; instead, “you develop it by actually doing the work.”
The conference marks 70 years since the 1956 summer workshop at Dartmouth where the term “artificial intelligence” was coined. These articles are being shared by partners in the Granite State News Collaborative; for more information, visit collaborativenh.org.
