How Students Perceive and Respond to GenAI for Peer Feedback Uptake

• In Event: Understanding and Supporting Feedback Uptake: Bridging Research and Practice in Educational Contexts

Abstract

1. Objectives
The rise of generative Artificial Intelligence (GenAI) has opened new avenues for scaffolding the peer feedback process (Sichterman et al., 2025). By offering targeted suggestions, clarifications, and strategic support in delivering and interpreting feedback, GenAI has the potential to facilitate meaningful peer feedback processes. Recent studies have focused primarily on exploring the role of GenAI for feedback provision rather than exploring how GenAI supports the uptake of peer feedback. This study explores how peer feedback uptake differs between students who integrate GenAI into their revision process and those who do not.

2. Theoretical framework
Peer feedback, a recognized instructional strategy with roots in Vygotsky’s sociocultural theory (Vygotsky, 1962), is considered instrumental in improving students’ learning (Topping, 2005). Peer feedback is a collaborative process of knowledge construction through the use of peer input (Kollar & Fischer, 2010). While GenAI support peer feedback provision, little attention has been given to their role in facilitating feedback uptake (Banihashem et al., 2024). Feedback uptake is a process by which students actively receive and integrate peer feedback (Alemdag & Yildirim, 2022). This paper is grounded in the social modes of co-construction in collaborative learning (Weinberger & Fischer, 2006) by exploring whether GenAI can scaffold these co-constructive processes during feedback uptake.

3. Method
Fifty Dutch graduate students participated in a three-week course with an online peer feedback activity. After receiving instruction, students wrote an essay, gave feedback to two peers, and revised their essays. Finally, A post-test questionnaire was provided to ask whether students used GenAI for peer feedback, dividing them into two groups: Tried AI (used at least one AI function; n = 15, 30%) and Never Tried AI (no AI use; n = 35, 70%) and asked students to rate their experiences, using a 4-point Likert scale.

4. Data sources and analysis
A coding scheme (see Gao et al., 2024) categorized peer feedback uptake into four types, namely Accept (n = 69), Elaborate (n = 153), Modify (n = 56), No uptake (n = 183), yielding 461 coded instances. Two raters independently evaluated uptake (κ = 0.73). A Mann-Whitney U test was conducted to examine group differences in uptake behaviors.

5. Results
Results show no significant differences in uptake behaviors between GenAI users and non-users across issue levels (Figure 1), suggesting limited impact of GenAI on uptake choices. However, users showed slightly more elaboration and modification.
Students who used GenAI found most functions moderately to very helpful (Figure 2), especially for verifying comment accuracy, identifying improvement areas, and assessing revision quality. However, functions like interpreting unclear feedback or suggesting revision methods were seen as only moderately helpful.

6. Significance
This study explores the overlooked role of GenAI in supporting peer feedback uptake. While results show no significant differences in uptake behaviors between users and non-users—suggesting GenAI alone may not drive meaningful co-construction of knowledge—users did show slightly more elaboration and modification. Although students mostly used GenAI for surface-level actions, these findings point to its potential to support deeper engagement.

Authors

• Omid Noroozi, Wageningen University and Research

• Golnoush Haddadian, Georgia State University

• Xingshi Gao, Wageningen University

• Christian D. Schunn, University of Pittsburgh

• Maryam Alqassab, Open Universiteit Netherlands

• Seyyed Kazem Banihashem, Open Universiteit Nederland