Chou, Yueh-Cheng
Lecturer, Department of Music, National Taiwan Normal University Taipei, Taiwan
E-mail：ken830615@gmail.com
Yeh, Jim PoTseng
Ph.D School of Journalism, Fudan University, Shanghai
E-mail：PTYeh22@m.fudan.edu.cn
Abstract
This study examines the integration of Generative AI and Scaffolding Theory into a popular music composition course at a comprehensive university, investigating its instructional design and practical outcomes. Adopting an action research approach, the study was conducted within a “Popular Music Composition” course at a comprehensive university, with eight graduate students of diverse disciplinary backgrounds as participants. A five-stage “Human-AI Co-creation” instructional module was developed—encompassing A&R concept development, characteristic lyric generation, phrasal deconstruction and reconstruction, adaptive melody streaming, and style generation with originality preservation—structured around a dual-scaffold system in which AI tools serve as “soft scaffolding” and instructor expertise as “hard scaffolding.” Results indicate that the dual-scaffold approach effectively reduced the cognitive load associated with technical execution, enabling all students to produce high-quality, industry-standard musical works; one student who fully applied the five-stage process successfully sold the resulting work to a record label, providing concrete evidence of academic-industry alignment. The study concludes that AI intervention does not replace the creator’s judgment; rather, through systematic scaffolding, it redirects students’ core competencies from “technical execution” toward “aesthetic curation and decision-making.” Human creators’ emotional intent and critical selectivity remain the inalienable foundation of creative agency—capacities that AI tools can amplify, but never supplant.
Keywords：popular music composition, generative artificial intelligence, scaffolding theory, human-AI co-creation, music education