This study investigates the impact of an AI tutor team member (AI) on students’ curiosity-driven engagement and learning effectiveness during an interactive molecular dynamics (IMD) task. We explore the role of the AI’s curiosity-inducing and -responsive behaviors in stimulating and sustaining students’ curiosity in an IMD task performed on the Visual Molecular Dynamics platform, as well as the impact on the frequency and complexity of student-led questions. We also evaluate how the AI intervention shapes student engagement, fosters exploratory curiosity, and enhances team performance within a learning environment. Using a Wizard-of-Oz paradigm, a human experimenter dynamically modulates the behavior of the AI tutor team member through a large-scale language model. Using a mixed-methods exploratory design, a total of 11 high school students participated in four IMD tasks (including molecular visualization and computation) of increasing complexity for 60 minutes. Team performance was assessed through real-time observation and video recording, and team communication was measured through the complexity of questions and the AI’s curiosity-inducing and -responsive behaviors. Cross Recurrence Quantification Analysis (CRQA) metrics reflect structural alignment of coordination and are linked to communication behaviors. High-performing teams demonstrated superior task completion, deeper understanding, and increased engagement. Advanced questions are associated with AI curiosity, indicating greater engagement and cognitive complexity. The CRQA metric emphasizes structured yet adaptive engagement to foster curiosity by highlighting dynamic synchronization in student-AI interactions. These proof-of-concept results demonstrate the ability of AI’s dual role as team member and educator to provide adaptive feedback and sustain engagement and cognitive curiosity.
