AI evidence extraction

The value of GenAI for peer feedback provision: student perceptions and impacts

Omid Noroozi, Golnoush Haddadian, Xingshi Gao, Christian D. Schunn, Maryam Alqassab, Seyyed Kazem Banihashem · International Journal of Educational Technology in Higher Education · 2025

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

8/10

Relevance

2/4

Quality (LMQS)

Evidence

Citations

97.65

FWCI

Top 10%

Impact

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.1186/s41239-025-00558-6

Methodology & findings

Study design

Mixed-methods observational study with survey (post-test questionnaire) and content analysis of peer feedback comments.

Sample

N = 50, 4 groups

Primary method

Descriptive analyses for Research Questions 1 and 2. Mann-Whitney U non-parametric tests for Research Question 3 to compare feedback features between groups (Tried GenAI vs Never Tried GenAI). Event sampling method for segmenting feedback into individual comments. Kappa (κ) reliability coefficients reported for coding reliability. Cronbach's alpha (α = 0.86) reported for overall helpfulness scale reliability. Software: ATLAS.ti for content coding.

Main result

The study found that "just over half of the students chose not to use GenAI for peer feedback provision, primarily because they believed they would learn more by completing the task independently." Among those who did use GenAI, "students who used GenAI provided more suggestions for high-level issues and offered less mitigating praise for low-level issues compared to those who did not use GenAI for peer feedback provision."

Reports effect sizes.

Research paradigm

Pragmatist/mixed methods (quantitative survey + coding analysis)

Author conclusions

"Together, these contributions advance the understanding of how GenAI can be used to support peer feedback provision effectively in higher education." The authors note that "utilizing GenAI within peer feedback setting in the form of so called 'hybrid intelligence' could promote collaboration between human (peers) and AI which could potentially contribute to the overall improvement of learning experiences." They conclude that GenAI can scaffold complex cognitive tasks but warn that "students need to be equipped with GenAI literacy skills to avoid misinterpreting information or overlooking potential inaccuracies and biases."

Risk of bias

Selection bias: Single institution, single discipline (food science), single course context; Attrition: 4 out of 54 students did not complete post-test questionnaire (92.6% completion rate); Confounding: No randomization; self-selection into GenAI use vs. no use; Social desirability bias: Students aware of instructor involvement and university AI policy may have biased responses; Measurement bias: Coders were not blind to GenAI use status during content analysis; Generalizability limitations: High proportion of female participants (86%), all English-language proficient graduate students; Single institution/context limits generalizability; Self-selection bias: students chose whether to use GenAI; Correlational design prevents causal inference; Single peer feedback activity may not capture long-term effects; Course instructor (first author) served as co-instructor, potential researcher bias; Attrition: 50 of 54 students completed questionnaire (92.6%); Selection bias: Participants were self-selected as they volunteered for the research part (mandatory attendance but voluntary research participation); Single institution bias: Study conducted at one Dutch university only; Attrition: 4 out of 54 students did not complete the post-test questionnaire; Single domain bias: Only food science graduate students; Social desirability bias: Participants may have reported perceived benefits of GenAI based on instructor guidance on ethical use; Correlational design: Cannot infer causality regarding GenAI effects

Limitations

"As an exploratory study, the relationship between GenAI use and peer feedback features was fundamentally correlational
A future experimental study is needed to test the causal influence of GenAI on peer feedback provision processes and outcomes
The study was also conducted within a single academic context, limiting the generalizability of the findings." Additionally, "This study was conducted within a single peer feedback activity, which might not capture the long-term impact of using GenAI on peer feedback provision."

Open questions raised

Need for experimental studies to test causal influence of GenAI on peer feedback provision
Research across diverse demographics, disciplines, and educational levels to improve generalizability
Long-term impacts of GenAI use on peer feedback provision (study captured only single feedback activity)
How to use GenAI without limiting learning opportunities for feedback providers
Effectiveness of GenAI for supporting higher-order writing skills beyond error correction
Role of instructors in guiding students to maximize GenAI's benefits (noted as underexplored)

Data: Not explicitly stated; data appears to be held by the authors; No datasets are mentioned as publicly available. The paper notes that email addresses used for data linkage were permanently removed after coding.Code: Not mentioned; Not mentioned in the paperExtracted from: pdfAgreement 47%

Explore related topics

AI Manuscript Writing Human-AI Collaboration Survey