Students perceive teacher feedback as more trustworthy than AI-generated feedback, but they value both for different, complementary purposes rather than seeing them as interchangeable.

Objective: This study investigated how university students use, value, and trust AI-generated feedback compared to feedback from educators, examining students' perceptions of both feedback sources and their impact on learning.

Methods: The researchers conducted a large-scale cross-sectional survey across four major Australian universities (Monash University, Deakin University, University of Queensland, and University of Technology Sydney). The study analyzed quantitative data from 6,960 respondents, including Likert-scale ratings of the helpfulness and trustworthiness of both feedback sources. Additionally, researchers performed thematic analysis on 8,642 open-ended responses to gain deeper insights into students' experiences. The survey was administered between August and October 2024, with participants representing diverse disciplines, study modes, and demographic backgrounds.

Key Findings:

• Nearly half of students (49.7%) reported using Generative AI (GenAI) for feedback on their academic work, while 50.3% did not.
• Both feedback sources were considered helpful by students, with 83.9% rating GenAI feedback as somewhat or very helpful and 82.2% giving similar ratings to teacher feedback.
• There was a significant difference in perceived trustworthiness: 90.5% of students rated teacher feedback as somewhat or very trustworthy, compared to only 60.1% for GenAI feedback.
• Students valued GenAI feedback for its accessibility (ease of access, speed, volume), understandability, objectivity, and low interpersonal risk compared to seeking feedback from teachers.
• Teacher feedback was perceived as more reliable, relevant, contextually appropriate, personal, and expert than GenAI feedback.
• Students who did not use GenAI for feedback cited concerns about trustworthiness (28.7%) and lack of awareness about this possibility (28.1%) as primary reasons.
• GenAI feedback was perceived to be more positive in tone, while teacher feedback was more likely to elicit negative feelings but was still trusted more.
• Students' emotional responses differed between the two feedback sources, with GenAI feedback perceived as less risky in an interpersonal sense but also less personally meaningful.

Implications: The study suggests that GenAI and teacher feedback serve different but complementary purposes in higher education. While GenAI offers immediate, accessible, and low-risk feedback that promotes understanding, teacher feedback provides more contextual, expert, and trustworthy information. Rather than replacing teachers, GenAI can complement their feedback, offering students additional opportunities to receive input on their work through different channels. Higher education institutions should consider how to support students in effectively utilizing both feedback sources, including developing students' evaluative judgment skills to critically assess AI-generated feedback.

Limitations: The researchers acknowledge several limitations. Despite the large sample size, the 5% response rate (about 10,000 students from an invited pool of approximately 192,000) may not be representative of the entire student population. Students with stronger views about GenAI may have been more inclined to participate. Additionally, the study was conducted in an Australian higher education context, which may differ from other regions. Finally, the survey was conducted at a time when few students would have experienced GenAI feedback explicitly integrated into their curriculum or assignments, which could change as universities provide more secure, customized GenAI tools.

Future Directions: The researchers suggest further investigation into how students process and implement GenAI feedback compared to teacher feedback. Key questions include: how students integrate different feedback sources, whether GenAI and teacher feedback continue to complement each other over time, how the timing of feedback affects its impact (with GenAI typically available before submission rather than after), and how students' perceptions of automated feedback may influence their engagement with it. The study also points to the need for research on how to develop students' feedback literacy in the context of GenAI.

Title and Authors: "Comparing Generative AI and teacher feedback: student perceptions of usefulness and trustworthiness" by Michael Henderson, Margaret Bearman, Jennifer Chung, Tim Fawns, Simon Buckingham Shum, Kelly E. Matthews, and Jimena de Mello Heredia.

Published On: May 13, 2025

Published By: Assessment & Evaluation in Higher Education (Taylor & Francis Group)