Beyond Answer Generation: University Teachers’ Experiences with a Reflective GenAI Chatbot for Professional Learning

Abstract

Generative artificial intelligence is increasingly used in higher education, yet limited research has examined how university teachers’ experience pedagogically designed AI systems that prioritize reflective learning over answer generation. This exploratory mixed-methods case study investigated university teachers’ experiences of a course-specific GenAI chatbot within a professional-development course at a Swedish university. All 48 course participants were invited to complete a questionnaire, of whom 21 responded and 11 reported using the course-specific chatbot; nine respondents also participated in semi-structured interviews. Questionnaire data were analyzed descriptively and interview data were analyzed using thematic analysis. The chatbot was designed to encourage dialogue, engagement with course literature, and independent thinking rather than provide summaries or ready-made answers. Participants generally found the chatbot easy to use, but perceptions of usefulness varied. Evaluations were shaped by prior experiences and expectations of GenAI: some teachers, who associated AI with efficiency and rapid answers, perceived the reflective design as restrictive, while others perceived its value as supporting dialogue and independent thinking. We interpret this tension as pedagogical friction, where design constraints intended to support reflection may conflict with users’ expectations of how GenAI should function. The findings suggest that the perceived educational value of GenAI depends not only on technical usability but also on alignment between pedagogical design, users’ expectations, and educational context. Successful implementation of reflective AI may therefore require explicit pedagogical framing alongside technical development.