ESSAY PROMPT

Use a recent student survey to discuss how AI use can affect classroom expectations and trust.

A gap between two answers

A student is writing about whether classmates feel pressured to use generative AI. A draft begins, “Everyone is using AI to cheat.” The sentence turns a suspicion into a fact and treats different kinds of use as identical. A more focused question is: why might students report less questionable use themselves than they attribute to classmates? Dartmouth released the first results of its 2026 AI Survey on 30 October 2026. Its findings can help explore that question, as long as the essay distinguishes reported behavior from beliefs about other people.

Tell readers which respondents count

The survey page reports a weighted national sample of 1,003 US college students identified through College Pulse. It also reports a separate Dartmouth sample, so a writer should not blend the two groups. The national sample is the relevant one for a statement about US college students broadly. “Weighted” describes how the results are adjusted for the intended population; it does not mean every student was asked or that the survey directly watched anyone use AI. Those distinctions belong in the setup because they determine what a percentage can represent.

In the margins Why this passage works +
“The national sample is the relevant one for a statement about US college students broadly.”

The sentence makes the scope decision explicit and prevents mixing a national sample with Dartmouth’s separate campus sample.

Compare reports with beliefs

One comparison makes the gap concrete. In the national sample, 8% reported copying and pasting AI text with little editing, while 50% believed their peers did. The two figures answer different questions: one asks about respondents’ own behavior; the other asks what they think other students do. The difference is evidence of a perception gap, not proof that the peer estimate is correct or that the self-report is complete. The survey page says responses were self-reported and notes that disapproval can contribute to underreporting of sensitive behavior. A careful writer therefore describes both the pattern and the uncertainty.

In the margins Why this passage works +
“The two figures answer different questions: one asks about respondents’ own behavior; the other asks what they think other students do.”

This distinction is the core reasoning move; percentages are not comparable as if they measured the same thing.

Use a question that keeps both sides visible

This pattern may affect classroom trust. If students think classmates use AI more than they admit, they may feel that ordinary take-home work is unfair or that they must keep up. But the survey alone cannot establish that the perceived gap caused any one student’s decision. It also cannot settle a universal rule about every tool or assignment. A better argument asks what teachers could clarify: which activities allow brainstorming, which require unaided work, and how students should disclose assistance. That proposal follows from the reported concern while remaining a recommendation rather than a result.

Keep concern and benefit in the same frame

The survey reports mixed views rather than a single verdict. Dartmouth says 84% of US college students surveyed believed AI use for writing can hurt students’ writing-skill development. That belief is important evidence about student concern, but it is not an experiment demonstrating that writing skill declined. Writers can use it to argue for explicit instruction in responsible assistance and independent practice. They should not convert a reported opinion into proof of an effect. The strongest thesis is narrower: students’ contrasting reports about themselves and peers suggest that clear, assignment-specific expectations could address uncertainty without assuming that every AI use is misconduct.

In the margins Why this passage works +
“That belief is important evidence about student concern, but it is not an experiment demonstrating that writing skill declined.”

The sentence separates an opinion measure from causal or performance evidence.

Separate measures before drawing a conclusion

When a survey asks what respondents do and what they think others do, keep those measures in separate clauses. Attribute self-reports, distinguish them from observed behavior, and avoid treating a perception as a verified fact about peers. Then propose a response as your own inference and explain the reasoning. Harvard’s guidance on analytical questions encourages writers to examine evidence and assumptions rather than settle for a broad topic. Here, the useful question is not simply whether AI is good or bad, but how mismatched expectations might affect fairness and learning.

Name the type of evidence

Overstated draft: “Half of students cheat with AI.” Revised: “In Dartmouth’s weighted national survey, 50% of respondents believed peers copied AI text with little editing, while 8% reported doing so themselves; both figures are self-reports, and the peer estimate is a belief rather than an observed count.” The revision preserves the striking gap while preventing it from becoming a false prevalence claim.

A CLOSER LOOK

The revision, explained

BEFORE

Half of college students copy AI answers.

AFTER

Half of Dartmouth’s weighted national sample believed peers copied AI text with little editing, while 8% reported doing so themselves.

The revision names whose belief the 50% represents and separates perception from self-reported conduct.

NOW IT’S YOUR TURN

Try the idea.

Rewrite “50% of college students copy AI answers” using the Dartmouth survey’s actual measure. Include the contrast with self-reported behavior and one limitation.

See the self-check +

One possible answer: “Half of the weighted national sample believed their peers copied AI text with little editing, while 8% said they did so themselves. The survey measured self-reports and beliefs, not observed copying, and sensitive behavior may be underreported.”

Sources and evidence

  • Dartmouth AI Survey 2026 · Dartmouth AI at Dartmouth
  • Asking Analytical Questions · Harvard College Writing Center