Beyond prompt engineering: the associations among creative self-efficacy, flow state, learner engagement, and psychological ownership in GenAI-assisted composition

Introduction: The integration of generative artificial intelligence (GenAI) into music education markedly lowers technical barriers for predominantly novice composers, but also raises concerns about a potential erosion of human creative agency. When learners rely on text prompts to produce music with minimal subsequent involvement, they may fail to develop a sense of psychological ownership over AI-assisted creations. Method: Drawing on the Theory of Psychological Ownership, this study examined the cognitive, affective, and behavioral processes through which perceived GenAI support relates to students' psychological ownership. Survey data were collected from 355 non-music-major undergraduates enrolled in a GenAI-assisted composition course that explicitly required iterative post-generation refinement of AI outputs. Structural equation modeling with bias-corrected bootstrapping was used to test a serial mediation model. Results and discussion: The results showed that perceived GenAI support was positively and significantly associated with psychological ownership, and that this relationship operated through a sequential pathway involving creative self-efficacy, flow state, and learner engagement. These findings suggest that GenAI does not inherently alienate learners; when positioned as a cognitive scaffold within a human-in-the-loop design, it is associated with creative confidence, optimal immersion, and active investment of effort. The study highlights the importance of deliberately incorporating productive friction into AI-supported learning activities to elicit an "IKEA effect," thereby transforming algorithmically generated material into personally appropriated creative artifacts.