Metacognitive Prompting for Responsible Generative AI Use: Learning Anxiety and Academic Self-Efficacy

This study examined whether metacognitive prompting for responsible generative AI (GenAI) use improves students’ learning anxiety and academic self-efficacy. A face-to-face classroom experiment (N = 148) compared a prompting condition with a control condition. Prompting increased metacognitive engagement (d = 1.23), reduced learning anxiety (b = −0.68, SE = 0.18, d = −0.63), and increased academic self-efficacy (b = 0.40, SE = 0.15, d = 0.44). Learning anxiety was negatively associated with self-efficacy (b = −0.42, SE = 0.06; r = −0.55), with an observed indirect pathway through anxiety (ab = 0.26). These findings suggest that brief metacognitive prompts can support responsible GenAI use while preserving learner agency, confidence, and psychological well-being.