A study has found a significant perception gap between faculty and students over how much university students use AI. Faculty judged students to be actively using AI, while students rated their own level of use as relatively low.

On Nov. 11, IT outlet TechRadar reported that joint researchers from the University of Manchester and the University of the Basque Country released findings comparing staff and student perceptions of generative AI use at universities.

The researchers surveyed 45 staff members and 86 students in the social sciences at a research-focused university in Britain from April to June 2024. The results were published in September in the international academic journal Frontiers in Education.

Staff rated students' AI use at an average of 5.23 out of 7. Students' self-reported score was 3.11, a gap of 2.12 points between the two groups.

Students mainly used AI to search for answers for assignments, write text, and handle translation and grammar correction. Students also showed a more positive attitude than staff that AI can help learning, reduce workload and raise productivity.

Staff experience with using AI varied. The share who had used AI at least once in the six months before the survey was 63%, but paid service subscribers accounted for 14%. The share of students with paid subscriptions was higher at 21%.

The researchers acknowledged that students may have reported using AI less than they actually did. At the same time, they explained that faculty also find it difficult to accurately grasp how students used AI based only on final assignment outputs.

The survey covered only some disciplines at a single university and the sample size was small. Because the data were collected in 2024, it is difficult to generalise the findings to the current state of AI use across universities as a whole.

Co-author Mark Carrigan (마크 캐리건), a senior lecturer at the University of Manchester, pointed out that discussions of AI in universities focus too heavily on concerns about cheating and the potential for educational innovation. He added that understanding how students actually use AI should come first. He stressed that universities should move away from responses based on suspicion and review teaching and assessment methods toward supporting learning.