Developing GenAI-Based Feedback Through Iterative Prompt Design: An Exploratory Study of Online Students’ Cognitive, Affective, and Behavioral Perceptions

This study reports the iterative prompt-engineering development of Generative Artificial Intelligence (GenAI)-based feedback in an online graduate course and examines students’ cognitive, affective, and behavioral perceptions. Grounded in the Student–Feedback Interaction Model, a qualitative case study involved 14 graduate students in a 14-week fully online course on open and distance learning. Students completed forum-based formative activities and received instructor-reviewed GenAI feedback. Informed by expert review, instructor observations, and student reflections, prompts evolved through three stages: checklist-style criterion verification, criteria-anchored synthesized explanations, and personalized diagnosis of strengths and weaknesses. Semi-structured interviews were analyzed through qualitative content analysis. Students reported improved feedback comprehensibility, detail, tone, and timeliness, valuing personalized, explanatory, criterion-based comments. However, inconsistencies between AI comments and instructor scores, limited conceptual evaluation, and lack of emotional presence weakened trust. GenAI therefore functions best as a draft-generation layer requiring sustained, active instructor oversight rather than as a stand-alone source of formative feedback.