Dartmouth’s first AI survey found that US college students are much more likely to think their peers use AI to cut corners than to say they do themselves.
The first results were released on 30 October 2026 at Dartmouth’s AI at 70 conference. The survey included 1,003 US college students in a weighted national sample and 1,533 Dartmouth students: 1,201 current undergraduates and 332 incoming students in the class of 2030. Dartmouth’s results are unweighted.
Students overestimate their peers’ use
For each of these tasks, students’ beliefs about peer use were higher than their own reports.
- 11% of US college students said they used AI to find problem set answers; 49% thought their peers did.
- 8% said they copied AI text with little editing; 50% thought their peers did.
- 23% said they used AI to draft essays; 55% thought their peers did.
These figures rely on self-reports, which can understate behaviour. The researchers explored possible underreporting by doubling students’ reported use. Large gaps between self-reported and perceived peer use remained for most tasks.
Students report pressure and fairness concerns
70% of US college students said take-home work had become unfair because of AI. 39% said they felt they had to use AI because their peers did.
Among incoming Dartmouth students surveyed before they arrived, 14% said they felt that pressure.
Most reported use supports learning
61% of US college students said they used AI to explain difficult concepts. 46% used it to make practice quizzes or study aids.
52% said working back and forth with AI produced better work than either they or the AI could produce alone.
Students also identified a risk to writing. 84% of US college students agreed that using AI for writing can harm the development of writing skills. Among Dartmouth undergraduates, 92% agreed, the highest agreement on any survey question.
Most students taught themselves
79% of US college students said they had mostly developed their AI skills on their own.
55% said strong AI skills would be important for their careers. At Dartmouth, 74% said the same. And 84% of Dartmouth students said AI is changing jobs in their field without necessarily eliminating them.
Questions for UK schools and colleges
The survey describes US college students. UK institutions need local evidence about how their learners use AI, and these findings can help frame questions for that work.
Do learners know what AI use is acceptable in take-home work? Are they overestimating how much their peers rely on it?
Dartmouth suggests that sharing reported use alongside clear guidance may ease pressure to keep up. The survey doesn’t test that approach in UK settings. Take-home assessment is one place for UK institutions to review how they explain acceptable AI use. What do your own learners think their peers are doing with AI?
The 79% self-taught figure also gives schools and colleges a reason to discuss AI skills before learners reach higher education.
For Digital Support Technician apprentices
If you’re working towards ST0120 version 1.1 or 2.0, use the survey as background when supporting colleagues with AI tools or contributing to acceptable use guidance. Note what you learned and how it informed relevant work. That short reflection may help explain your learning in your portfolio.