Investigating student perceptions of an AI-powered chatbot to support feedback interpretation and uptake in higher education

Assessment feedback is widely recognised as a powerful influence on learning, yet students often struggle to act on comments they receive. While much research examines AI systems that generate feedback, less attention has been paid to tools supporting engagement with feedback already provided. This study investigates MSc computing students’ perceptions of a proposed AI-powered chatbot for supporting feedback engagement within an online postgraduate programme at a UK research-intensive university. Rather than evaluating a deployed tool, the study explored how students envisaged a chatbot functioning as a dialogic mediator grounded in their submissions and markers’ comments. Data were collected through four focus groups and a structured survey. Thematic analysis revealed disengagement was shaped by structural and affective barriers, including delayed feedback, inconsistent standards, and the social cost of seeking clarification; 69% of respondents wanted to ask questions but did not. Students valued Socratic, dialogic interaction over direct answer provision, saw contextual grounding as essential for trust, and maintained human judgement as a non-negotiable boundary. All respondents rated the tool as at least slightly useful. The findings contribute to feedback literacy and AI in education scholarship by conceptualising AI as a feedback interpreter and offer design guidance for institutionally governed feedback-support tools.