Student Perception of AI-Supported Feedback Systems in Higher Education

Sammanfattning

This study examines how Higher Education (HE) students perceive AI feedback systemsacross three distinct conceptual configurations, the Fixer, Coach, and Mentor, and identifiesthe student-side and institutional conditions that enable or constrain those perceptions. Grounded in Affordance-Actualisation Theory (AAT) and Hattie and Timperley's four-level feedback framework, the study adopts an interpretivist, qualitative approach using scenario-based vignettes embedded within six focus group sessions with twentyparticipants across disciplines and study levels. Findings reveal that perception variessystematically across configurations: the Fixer is universally recognised as feedback butperceived as limiting deeper learning, while the Coach and Mentor are evaluatedconditionally, mediated by students' prior feedback conceptions, subject knowledge, andmetacognitive capacity. Time pressure emerged as the dominant institutional conditionconstraining access to higher-order affordances. Two theoretical extensions to AAT areproposed: longitudinal affordance foreclosure, whereby repeated Fixer over-useprogressively erodes student-side facilitating conditions, and an evaluative dimension ofaffordance perception, through which students' prior feedback conceptions function as apre-perceptual filter. An emergent finding establishes human feedback as the relationalstandard against which AI configurations are evaluated.