Think–Pair–Chatbot–Share: AI-Facilitated Peer Learning in Chemistry

Abstract

This study introduces and evaluates the Think–Pair–Chatbot–Share (TPCS) model, an adaptation of the traditional Think–Pair–Share (TPS) framework where a generative AI (Gen-AI) chatbot is integrated to scaffold learning and provide feedback. We compared first-year chemistry students learning ionization energy and atomic structure using traditional TPS versus the Gen-AI-facilitated TPCS model. Student responses from both think and share stages were analyzed for accuracy and common errors, while engagement was examined via participation rates. Results indicate that for students who completed all stages, the Gen-AI-facilitated TPCS group recorded a higher proportion of correct share stage responses in three of the four tasks. However, the Gen-AI-facilitated TPCS intervention was associated with lower overall participation, suggesting potential challenges related to sustained engagement with the Gen-AI-facilitated, multitask activity. These findings offer the first classroom-level evidence of AI-mediated TPS in action within tertiary chemistry education and underscore both its instructional potential and the challenges in maintaining engagement.