Exploring cognitive load dynamics with an AI-based voice assistant in high school chemistry experiments

Laboratory courses in chemistry education inherently involve cognitive and sensorimotor tasks that demand substantial cognitive resources. Using a Directed Acyclical Graph (DAG) approach coupled with Bayesian multi-level modeling, we investigated the impact of an AI-driven voice assistant system on high school students‘ (N = 55) prior knowledge (PK) and intrinsic cognitive load (ICL)’ during an acid-base titration experiment in a real classroom setting. The voice assistant, specifically designed for lab environments, aims to reduce cognitive overload via hands-free, step-by-step instructional support, thus potentially reducing the effect of switching back and forth between lab work and instruction on cognitive load, leveraging students’ prior knowledge. While synthetic data simulations based on theoretical assumptions confirmed a negative relationship between PK and ICL, empirical data revealed more variability and an unexpected positive association. Notably, participants experienced in lab settings consistently reported lower cognitive loads, suggesting that prior practical exposure significantly influences how technology-assisted instructions impact cognitive load. The findings underscore the necessity of considering institutional contexts, instructional complexity, and individual differences in practical experience when integrating advanced digital tools into chemistry education. Future research should expand on these insights, exploring diverse experimental contexts and larger student populations to optimize the integration of AI-driven assistance in educational laboratories.