ABSTRACT

This study aimed to explore the characteristics of interaction and perceptions of collaboration with AI that emerge when composition majors and non-majors use AI-based music generation tools during the composition process. Twenty Korean undergraduates (10 composition majors, 10 non-majors) completed an individual composition task using Soundraw, with data collected through surveys, screen recordings with think-aloud protocols, and post-task interviews. Results showed that composition majors tended to engage in detailed modifications and refinements of the generated outputs, whereas non-majors more frequently relied on repeated playback and regeneration. In addition, non-majors were more likely to regard AI outputs as finalized results, while composition majors viewed them as flexible creative materials that could be further modified and developed. Although both groups acknowledged AI’s efficiency and idea-generating benefits, their educational needs for AI use differed according to their level of musical understanding, highlighting the need for learner-level-based and personalised music education designs in which AI is positioned as a collaborative partner requiring critical understanding and creative utilization.

Keyword :