#1110: Exploring Learner Engagement in an LLM-Supported Simulation Environment for Complex Systems Understanding

Understanding complex systems is challenging, as learners often rely on linear reasoning and struggle to recognize decentralized and emergent processes. Large language models (LLMs) offer new opportunities to support inquiry in simulation-based learning, yet how learners engage with these tools and how such engagement relates to learning outcomes remains underexplored. In this study, we investigate learner engagement within a theory-informed, LLM-supported simulation environment, based on an experiment with 69 postsecondary students that varied parameter control and questioning agency. Focusing on questioning agency, we examine how interactions with an LLM-based tutor and engagement with exploratory instructional regions relate to posttest performance. Results show that greater engagement with the tutor, particularly through content-focused interactions, is associated with higher learning gains. Time spent in exploratory regions also significantly predicts performance. These findings highlight the importance of fostering meaningful engagement with both conversational tutors and interactive features in LLM-supported environments.