Dartmouth survey finds students see more AI shortcuts in peers than themselves

College students describe their own AI use as help with learning, but view classmates as far more likely to take shortcuts. Dartmouth released that finding on October 30, 2026, alongside a measure of the tension around homework: 70% of surveyed U.S. college students said AI had made take-home work unfair.

The inaugural Dartmouth AI Survey gathered responses between June and September from more than 1,500 Dartmouth students and 1,000 college students across the United States. Economics professor Bruce Sacerdote leads the multi-year project. Its first findings were presented at Dartmouth’s AI at 70 conference, offering a picture of students’ habits, suspicions and expectations about work.

A wide gap in students’ accounts

Only 11% of college students reported using AI to solve problem sets, while 49% believed their peers did so. The difference was larger for copying and pasting AI-generated text with little editing: 8% reported doing it themselves, but 50% believed their peers did.

Dartmouth students showed a similar divide on essay writing. Approximately 17% of undergraduates said they used AI to draft essays, while they believed 63% of their peers did. These comparisons put students’ accounts of themselves beside their impressions of others; they do not establish which account better reflects actual behavior.

The fairness concern also came with pressure to participate. Thirty-nine percent of surveyed U.S. college students said they felt pushed to use AI because their peers did. Sacerdote connected those assumptions about classmates to how students judge fairness and try to keep up.

Students have told us they want to use AI to learn and prepare for their future careers, but at the same time, they assume the people around them are cutting corners.

Bruce Sacerdote, Dartmouth economics professor and survey research lead

Useful help, feared skill loss

Students’ concerns did not amount to a rejection of AI as a study tool. Nationally, 61% said they used it to explain difficult concepts. Another finding focused on collaboration: 52% said back-and-forth use of AI produced better work than working on their own.

Those perceived benefits sat alongside widespread concern about writing. Eighty-four percent of U.S. students said using AI for writing could hurt the development of writing skills. At Dartmouth, the share was 92%. Both findings concern what students believe AI can do, rather than a test of whether their skills improved or declined.

Views also varied by subject at Dartmouth. Seventy percent of science students said AI deepened understanding, compared with 67% in social sciences and 48% in arts and humanities. The survey found that differences by field of study were twice as large as differences by class year.

Preparing for work, mostly by themselves

For many respondents, learning to use AI had been largely an individual effort. Seventy-nine percent of surveyed U.S. college students said they had mostly taught themselves their AI skills. Dartmouth described students as arriving at college with little AI guidance from high school.

Career expectations differed between the Dartmouth and national groups. Seventy-four percent of Dartmouth students considered AI skills critical to their careers, compared with 55% nationally. Among Dartmouth students, 84% said AI was changing jobs in their field, but not necessarily eliminating them—a view of changing work rather than inevitable job loss.

The project’s next phase will examine employers’ perspectives, too. Dartmouth is surveying employers about whether entry-level jobs are growing or shrinking, and how AI is changing required responsibilities and skills. The next student survey is scheduled for August 2027.