Negotiated Creative Agency in AI-Assisted Music Production: A Grounded Theory Study of Independent Musicians

Abstract

This study addressed the unresolved problem of how independent musicians preserve, redistribute, and legitimate creative agency when generative artificial intelligence can produce musically coherent lyrics, melodies, vocals, arrangements, and complete songs. It pursued three objectives: (1) to examine how independent musicians engage with AI tools in music-production workflows; (2) to explain how AI-assisted production reshapes their perceptions of creativity, creative control, and authorship; and (3) to develop a grounded theoretical model of how creative agency and authorship are negotiated across different forms of AI-assisted production. A qualitative grounded-theory design was used with 30 independent musicians selected through purposive and theoretical sampling. Data were generated through semi-structured interviews lasting approximately 45–60 minutes and analyzed through constant comparison, open coding, axial coding, selective coding, and analytic memo writing. The results found that: (1) musicians primarily used AI to accelerate ideation, overcome creative blocks, test genres, and produce rapid drafts, but meaningful use required repeated prompting, selection, editing, and recombination; (2) technical quality alone did not establish creativity or authorship, because participants distinguished outputs that merely “sounded good” from those that “sounded like me,” making aesthetic evaluation, identity alignment, control restoration, and contribution assessment central to the process; and (3) negotiated creative agency emerged as the core category, producing four modes—tool-assisted creation, human-led co-creation, hybrid authorship, and AI-dominant production. The study extends distributed-agency and human–AI co-creation research by showing that agency is not simply shared; it is recursively evaluated, reclaimed, surrendered, and legitimated.