Academic socialisation through AI and peer feedback in doctoral writing: information and orientation in feedback ecologies

Journal article, 2026

Academic socialisation in doctoral writing is shaped through feedback that helps writers understand how research is argued, positioned, and read within a community. This study examines how Generative AI and peer feedback contribute differently to doctoral students’ academic socialisation when working on authentic, high-stakes texts for publication. Forty STEM doctoral students in a Writing for Publication course drafted an introduction to their research article, received written peer feedback, AI feedback, and submitted reflective critical analyses comparing the two sources. Using content analysis of 954 feedback moves and thematic analysis of 40 reflections, the study identifies clear functional distinctions. AI produced abundant, explicit, directive task-level feedback that efficiently improved textual features, yet offered limited epistemic, rhetorical, or audience-oriented guidance. Peer feedback, although narrower in scope and often implicit, provided richer cues about rationale, disciplinary expectations, and audience interpretation. Students valued AI for its speed and actionability but trusted peer feedback more, viewing it as more credible, transferable, and aligned with the developmental trajectory of becoming a researcher. The findings underline the complementary yet non-equivalent roles of AI and peers; AI primarily provides information for revising the text, whereas peers provide orientation to disciplinary norms and audience expectations that supports academic socialisation.