One in Four Surveyed Students Had Used University AI Guidance

A policy on the LMS is easy to point to. In a tutorial, the more useful question is whether students can apply it: what may they use AI for, what must they disclose, and which claims should they check?

A survey of accounting and finance students suggests that familiarity with generative AI should not be mistaken for engagement with university guidance.

What the survey asked

Vangelis Tsiligkiris, Anita Kéri and Dorothea Bowyer surveyed 523 students in the first half of 2024, at one university each in the UK, Australia and Hungary. Students answered rated questions and open questions about their use of GenAI.

Familiar tools, less-used guidance

About 73% said they were at least moderately familiar with GenAI tools. Only 24% reported that their university had AI guidelines and that they had used them.

The researchers also found an association between students' responses about the existence and use of institutional guidelines and how often they said they checked AI-generated content. That is a useful lead for teaching, but it is not a test showing that a policy changes behaviour.

For lecturers, the practical distinction is between publishing a rule and helping students use it in a real task. The authors' teaching proposals are illustrative, rather than evaluated interventions.

Limits and context

The results are self-reported and come from a single snapshot. Each country is represented by one institution, with unequal response numbers, so these are not national comparisons. Volunteers may differ from other students. The survey cannot establish whether guidance changed checking habits or whether students who checked more were more likely to engage with guidance.

Evidence strength: early signal. This is one survey of self-reported behaviour at three institutions, taken once. It shows an association between using guidelines and checking AI output, not proof that guidelines change behaviour.