#647: Students’ Mental Models of AI Chatbots in Inquiry Learning: a Friend, an Instructor, a Browser, and a Partner

This study investigates how high school students perceive and interact with an AI chatbot during a science lab experiment. Analysis of chatbot transcripts revealed four enacted roles in student-AI interaction (Instructor, Browser, Human, and Reflective Partner) that reflect students’ underlying mental models and how they conceptualize the chatbot’s capabilities and relationship. Students’ conversations shifted across these roles, with Browser and Instructor predominating. In contrast, post-activity surveys offered retrospective perspectives that largely framed the chatbot as an Instructor and, less frequently, as a Browser, overlooking the reflective and human-like engagements evident in transcripts. By confronting students’ talk about AI with their live interactions during inquiry, this study highlights mental models as dynamic, enacted in practice, and locally and culturally constructed in conversation, suggesting directions for designing educational AI systems that can accommodate dynamic roles.