Can machines truly create music? Toward a redefinition of creativity in the age of generative AI

Abstract

Generative artificial intelligence (AI) is now capable of producing musically coherent audio from short prompts, reviving old philosophical questions regarding creativity, authorship, and the distinctiveness of human expression. This article develops a conceptual analysis of AI-generated music as a sociotechnical practice that reshapes not only ideas of creativity and authorship, but also the material conditions under which human artists work. After tracing how Western aesthetics progressed from divine inspiration and imitation to Romantic originality and the twentieth-century valorization of innovation, this study situates contemporary systems within the paradigm of computational creativity. Drawing on accounts that frame creativity as recombination and search within structured spaces, this article explains why generative models can yield outputs that listeners experience as novel and expressive. Simultaneously, intentionalist and contextualist perspectives highlight the deficiencies of current systems: lived experience, purposive self-expression, and situated participation in cultural worlds. The core argument is that musical meaning and value are not properties of sound alone, but emerge through distributed authorship involving model designers, training data ecologies, prompt writers, performers, platforms, and listeners. Rather than asking whether machines “feel,” this article reframes the question as “Can machines truly create music?” to investigate how responsibility, consent, attribution, and accountability are allocated across the generative stack of technologies. It concludes by outlining practical norms for human–machine musicianship, including transparency, consent-sensitive data practices, and contextual attribution, which allow AI to expand creative practices without eroding artistic agency or audience trust.