From cognitive relief to motivational enhancement: associations between perceived Agentic AI Capability and student engagement in systems development courses

Generative AI agents are increasingly emerging as important learning partners for supporting complex problem-solving in higher education; however, the psychological processes associated with student engagement remain insufficiently understood. Grounded in Cognitive Load Theory and Social Cognitive Theory, this study examines how perceived Agentic AI Capability is associated with students' extraneous cognitive load, academic self-efficacy, and overall student engagement in systems development courses. An empirical study was conducted among undergraduate students majoring in e-commerce who participated in systems development-related courses. The results indicate that perceived Agentic AI Capability is negatively associated with extraneous cognitive load and positively associated with academic self-efficacy. Extraneous cognitive load is negatively associated with academic self-efficacy and student engagement, whereas academic self-efficacy is positively associated with student engagement. The indirect associations through academic self-efficacy and the sequential pathway are statistically significant. The findings provide evidence on cognitive and motivational correlates of AI-supported learning while, given the cross-sectional design, not establishing causal or temporal relationships. This study extends the application of Cognitive Load Theory and Social Cognitive Theory to human–AI collaborative learning environments and provides theoretical insights and practical implications for integrating generative AI into systems development education.