Dartmouth AI Survey 2026

About the Survey

A multi-year study of how AI is shaping learning and career readiness for U.S. college students and Dartmouth undergraduates.

The first results were released on Oct. 30 at Dartmouth’s AI at 70: What Must Remain Human? conference.

Lead researcher: Bruce Sacerdote ’90, professor of economics, Dartmouth
Coauthor: Mason Parris-Bacon, Economics research specialist, Dartmouth

Explore the Dashboard

Read the Working Paper

Survey at a Glance

In the 2026 Dartmouth AI Survey, few students say they use AI to cut corners, but many think their peers do. Most of them report using AI to learn, and they see its benefits and its risks. Many also say take-home work now feels unfair, that they are mostly self-taught in AI, and that AI skills will matter for their careers—though they don’t believe AI will cause a “jobpocalypse.”

Survey Sample

1,003

college students around the United States, weighted national sample

(identified through College Pulse)

1,533

Dartmouth students

(current undergraduates and the incoming class of 2030)

The survey will repeat yearly, and a prospective employer survey is under way.

Few students say they use AI to cut corners, but many think their peers do.

Many say take-home work has become unfair because of AI. Some feel pressure to use it.

Students see AI’s potential harm to learning, but they also see real benefit.

84% of U.S. college students say using AI for writing can hurt the development of students’ writing skills
61% of U.S. college students use AI to explain difficult concepts
52% of U.S. college students say working back and forth with AI produces better than either they or AI would alone
At Dartmouth, 92% of undergraduates say that AI can hurt writing skills, the highest agreement on any question in the survey, while 66% use AI to explain difficult concepts.

Students arrive with little AI guidance from high school, and most teach themselves.

Views differ more by field of study than by class year.

Students believe AI skills will matter for their careers, but they don’t believe in a “jobpocalypse.”

What this Means for Colleges and Universities

Lift the veil on AI use. U.S. college students believe their peers use AI far more than they report using it themselves. Sharing what students report, along with clear guidance on appropriate uses and academic integrity, may alleviate the pressure many feel to keep up.
Rethink how students learn and are assessed. Seven in 10 U.S. college students say take-home work has become unfair because of AI. There’s no single solution, but assignments and course policies may need to change and be discipline-specific.
Plan for students arriving self-taught. Most U.S. college students say they developed their AI skills on their own, and many enter college with little guidance from high school.

“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. This shapes how they feel about fairness and how they work to keep up. The good news is that we can do something about this. Universities can help students think critically and contribute meaningfully even in the age of AI.”

Bruce Sacerdote ’90, professor of economics

Survey Details

National sample: 1,003 U.S. college students, identified through College Pulse, a company founded by Dartmouth alumni. Results are weighted to be nationally representative. September 10–23, 2026
Dartmouth undergraduates: 1,201 current undergraduates responded, about 25% of the undergraduate student body, and 332 incoming students in the Class of 2030. The survey was fielded beginning June 8, 2026.
Questions: Dartmouth results are unweighted.
Self-reported: Answers are self-reported. The students were first asked about their peers before themselves.

About the Researchers

Bruce Sacerdote '90 is the Richard S. Braddock 1963 Professor in Economics at Dartmouth. He enjoys working with detailed data to enhance understanding of why children and youth turn out the way they do. He is also involved in a series of studies to examine how students make choices about going to college and how policymakers might influence that decision-making process.

Mason Parris-Bacon is a pre-doctoral researcher at Dartmouth. He is headed to a PhD program in economics next year. He is a Special Sworn Status researcher with the U.S. Census and separately works with Internal Revenue Service Statistic of Income under the Interagency Personnel Agreement. His fields include Public Economics and the Economics of Higher Education.

How AI was used

Survey and interviews: The national sample of 1,003 U.S. college students was identified through College Pulse’s panel of college students. Interviews with prospective employers are conducted by a new AI interviewer tool led by MIT.
The Analysis: AI helped write the code for the survey dashboard, which is the source of the figures in this page and in the slides. AI was used to analyze open-ended answers to look for themes. Those answers were coded with keyword rules, and the final screening was done by hand. All dashboard figures were checked against the raw survey data.

What is the Dartmouth AI Survey?

Who took the survey?

How was the national sample selected?

How do college students use AI?

How do students think their peers are using AI?

Do students report their own AI use accurately?

Probably not perfectly. Like any behavior survey, the Dartmouth AI Survey relies on what students say they do. This can lead to underreporting of behaviors they think others might disapprove of. The survey was anonymous and intentionally asked about specific uses of AI rather than breaking rules to help reduce the effect.

To explore how much underreporting could have impacted the results, the research team, guided by research on underreporting on sensitive topics, doubled student’s self-reports. Even after this, large gaps remained between self-reported use and beliefs about peer use for most uses. While the survey cannot say exactly how much students use AI, it does show that students believe their peers use it far more than they report using it themselves and that many say they feel pressure to use it to keep up.

Do students think AI hurts learning?

Will the survey be repeated?

AI at Dartmouth