A survey of US college students found a striking gap between how often respondents said they use AI and how often they think their peers do. Half believed classmates copy and paste AI text with little editing, while 8% said they did so themselves, according to a Psychology Today account of the 2026 Dartmouth AI Survey.
Watch Desk analysis
What happened
The survey covered 1,000 US college students and more than 1,500 Dartmouth students. Among US respondents, 39% said they use AI to summarise assigned readings, while 60% believed their peers do. For using AI to find problem-set answers, the figures were 11% and 49%, respectively. The reported results also show that 23% said they use AI to draft essays or papers, compared with 55% who thought their peers did.
The article attributes the survey to Dartmouth economics professor Bruce Sacerdote and co-author Mason Parris-Bacon. They suggest the gap may reflect both reluctance to admit AI use and students overestimating how much their peers use it. The survey is an inaugural release in a planned multi-year study of AI’s effects on learning and career readiness.
Key findings
- Peer use looks higher than personal use
Across several tasks, students’ estimates of classmates’ AI use substantially exceeded their own reported use. - Fairness is part of the picture
70% of US respondents viewed AI use as making take-home work unfair; 39% said they had used AI because their peers did. - Students see career stakes on both sides
55% said strong AI skills would matter for career success, while 56% thought AI was reducing opportunities in their desired field.
Why it matters
The findings point to a practical challenge for universities: students may be responding not only to what AI can do, but to what they believe everyone else is already doing. If take-home work feels unfair, changing rules for assessment may matter as much as issuing another campus-wide reminder to use tools responsibly.
The gap between self-reports and estimates of peer behaviour is not proof that students are hiding their own use. The researchers’ account, as summarised by Psychology Today, offers two possible explanations: under-reporting and overestimating peers. Either way, perceptions themselves could shape classroom norms.
Our read
The most useful signal here is not that students have discovered AI. It is that they may see its use as more widespread than their own answers suggest, and many already regard take-home work as unfair. Universities need to understand that perceived arms race before designing rules around it. A survey can map the problem; it cannot settle what good assessment looks like.
What to watch
- Whether later waves of the study find that self-reported use or perceptions of peer use change.
- How universities adapt take-home assessment and AI guidance.
- Whether the reported gap appears across subjects and institutions, not just in these samples.
Discussion spark: If students think classmates are using AI more than they are, should universities redesign take-home work around that perception, or focus first on clearer rules and enforcement?
Sources and evidence
- Source update (30 October 2026, 23:18 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.