As institutions of higher education grapple with how to integrate generative AI into assessment, our open-access article in Technology, Pedagogy and Education examines how university students themselves experienced AI-generated feedback alongside anonymised peer review on substantial, open-ended coursework. Drawing on linguistic reflections and multimodal artefacts (combining images, gesture, and space for meaning-making) produced by 93 educators enrolled in two online postgraduate courses at a public Midwestern university in the United Sates, we used thematic and social semiotic analyses to uncover our participants’ views on benefits, differences, and limitations of each feedback source.
Across the participating cohort, students clearly preferred human peer reviews over their AI counterparts for their specificity, contextualisation, and interpersonal support. Peers were seen as engaging with argument, structure, and style in more actionable ways. Additionally, our participants felt that reviewing their classmates’ work resulted in improved self-revision. Despite this preference, the students in our work also seemed to value the AI reviewer’s speed, consistency, and rubric alignment, especially for early, “big-picture” checks. Taken together, the two feedback types mapped productively onto Hattie and Timperley’s (2007) model: AI feedback clustered at the task/criteria level (Where am I going?), while peer comments addressed process and self-regulation (How am I going? and Where to next?). The practical implication is not AI versus peers, but the thoughtful orchestration of both feedback types, achieved through calibrated prompts tied to transparent rubrics and human dialogue that attends to meaning, audience, and voice.
Our conclusion is straightforward: When well-designed and embedded in an explicit pedagogy (in our work, Learning by Design), generative AI can complement, not replace, peer reviews to create richer feedback ecologies. For instructors, this might suggest sequencing AI for rapid, criteria-based diagnostics and reserving peer/human time for deeper interpretive work and encouragement. For students, it might legitimise AI as a useful (albeit limited) partner in revision.
Explore the full study and its methodological details here (open access): https://doi.org/10.1080/1475939X.2025.2480807.
Citation: Zapata, G. C., Cope, B., Kalantzis, M., Tzirides, A. O, Saini, A. K., Searsmith, D., Whiting, J., Kastania, N. P., Castro, V., Kourkoulou, T, Jones, J., & Abrantes da Silva, R. (2025). AI and peer reviews in higher education: Students’ multimodal views on benefits, differences and limitations. Technology, Pedagogy and Education, 1-19. https://doi.org/10.1080/1475939X.2025.2480807
Gabriela C. Zapata
Gabriela C. Zapata holds a PhD in Spanish (Linguistics track) from the Pennsylvania State University. She is Associate Professor in Education at the University of Nottingham. She also serves as the editor of the book series Multiliteracies and Second Language Education (Routledge) and co-editor-in-chief of the journal Diversity & Inclusion Research (Wiley). Her main research areas are Generative AI in higher education instruction, assessment, and teacher training; AI literacy; multiliteracies-based instruction (focus on Learning by Design); and multimodal social semiotics. Throughout her career, she has published articles in a variety of peer-reviewed journals and edited volumes, as well as four books on multiliteracies-based second/heritage language education and four textbooks for the teaching of Spanish as a second language. She has also been involved in the development and implementation of inclusive, research-guided methodologies and open educational resources for language teaching and teacher education. Dr. Zapata has been part of several interdisciplinary collaborative projects that have had the objective of serving Hispanic/Latinx and Black/African American students and communities in Southern United States and the Salinas Valley in California. Her latest edited volume explores the role of AI technologies in multiliteracies-based language education (Routledge), while her forthcoming co-authored book (with Drs. Bill Cope and Mary Kalantzis) focuses on literacy development in the age of Generative AI.
