AI or Human? Evaluating Student Feedback Perceptions in Higher Education

Henderson et al. (2025) and Nazaretsky et al. (2024) found that knowing feedback was AI-generated may reduce trust compared to teacher feedback, making students more cautious in acting on it.

“Knowing it was AI-generated may have reduced trust compared to teacher feedback, making students more cautious in acting on it (Henderson et al., 2025; Nazaretsky et al., 2024).”

Due to limited tutor resources and increasing student-teacher ratios, feedback often becomes generic and delayed, failing to adequately meet students' individualised learning needs.

“Due to limited tutor resources and increasing student-teacher ratios, feedback often becomes generic and delayed, failing to adequately meet students' individualised learning needs (Dai et al., 2024; Meyer et al., 2024; Nazaretsky et al., 2024).”

GenAI should not be viewed as indisputably superior, but rather as a feasible and scalable complement to human feedback.

“Accordingly, GenAI should not be viewed as indisputably superior, but rather as a feasible and scalable complement to human feedback (Henderson et al., 2025; Nazaretsky et al., 2024; Zhang et al., 2025).”

Students often attribute greater genuineness to GenAI-generated feedback, especially under blind conditions.

“this finding aligns with previous work by Nazaretsky et al. (2024), who also observed that students often attribute greater genuineness to GenAI-generated feedback, especially under blind conditions.”

Understanding students' perceptions is of particular significance, as these perceptions shape not only students' engagement with feedback but also the extent to which they interpret and act upon it, ultimately influencing their overall satisfaction with the learning process.

“In addition, understanding students' perceptions is of particular significance, as these perceptions, as conceptualised by Nazaretsky et al. (2024), encompassing objectivity, usefulness, genuineness, and credibility of the provider, shape not only students' engagement with feedback but also the extent to which they interpret and act upon it, ultimately influencing their overall satisfaction with”

The variation in students' ability to identify whether feedback was AI- or tutor-generated aligns with findings from Turing test-like designs.

“This pattern aligns with findings from studies employing Turing test-like designs, which similarly observed variation in students' ability to identify whether feedback was AI- or tutor-generated (Nazaretsky et al., 2024; Zhang et al., 2025).”

Findings are inconsistent about whether recipients can tell if the feedback provider is AI or human.

“Findings are also inconsistent about whether recipients can tell if the feedback provider is AI or human, ranging from studies reporting that people could distinguish it reasonably well [33] to studies reporting that they could hardly distinguish it [3435].”