AI as a dialogic partner: rethinking feedback in higher education

There is a well-established body of research on effective feedback practices in higher education. Yet, student engagement with feedback remains varied. One possible explanation lies in limited feedback literacy, particularly the difficulty students face in managing the affective dimensions of receiving critique. Emotional responses can inhibit students from acting on lecturer feedback. This raises the question: can removing the relational aspect of feedback improve engagement? Generative AI (GenAI) tools are increasingly embedded in various educational contexts, supporting both students and teachers by offering real-time feedback and facilitating assessment processes. However, integrating GenAI into feedback practices in higher education requires a paradigm shift, one that positions GenAI not as a threat or all-knowing authority, but as a collaborative learning partner with the potential to foster dialogic engagement with students. With this in mind, we adopted a Bakhtinian approach to dialogue to qualitatively explore whether AI-generated feedback can reshape students’ engagement by reducing affective barriers and legitimising AI as a dialogic partner in learning. Drawing on Bakhtin’s concepts of authoritative and internally persuasive discourse, we investigate whether interaction with GenAI feedback encourages greater student agency and engagement with the feedback process among students from widening participation backgrounds at a London post-1992 university.