This study aims to explore the impact of varying levels of cognitive load on students’ learning behaviors when using ChatGPT in generative AI-assisted instructional settings. Grounded in Cognitive Load Theory, the research involves the design of an instructional module on “Systems Analysis and Design” within the field of Information Management. An experimental approach was employed to examine the mechanisms through which ChatGPT influences extraneous cognitive load, intrinsic cognitive load, and students’ continued intention to use the tool. It is hypothesized that both intrinsic and extraneous cognitive load experienced by students in ChatGPT-assisted learning environments negatively affect their willingness to continue using the tool. To test this hypothesis, Structural Equation Modeling was utilized for data analysis. The results reveal that both intrinsic and extraneous cognitive load exert significant negative effects on continued usage intention. This suggests that excessive cognitive load during the learning process may reduce students’ willingness to adopt ChatGPT as a learning tool in the future. Overall, this study provides both theoretical and practical implications for the integration and design of generative AI tools in educational settings.