CCMSAI 2026

Within the Euro-American art-music tradition, creativity has largely been framed through humanist paradigms privileging individual authorship, intentionality, and stylistic innovation. The growing presence of generative AI and machine learning (ML) in composition and performance unsettles these assumptions and calls for a rethinking of creative agency. This paper asks how the integration of artificial neural networks into live vocal improvisation reshapes musicological concepts of authorship, performance, and agency. Drawing on Donna Haraway’s situated, hybrid subjectivities, Rosi Braidotti’s relational posthuman subject, Deleuze and Guattari’s theory of assemblage, and Dominic Pettman’s account of sonic intimacy, it develops a posthumanist framework that is tested against two case studies: Tomomibot, by Tomomi Adachi, Andreas Dzialocha, and Marcello Lussana, and ULTRACHUNK, by Jennifer Walshe and Memo Akten. Combining philosophical inquiry with musicological analysis of vocal technique, performer-interface interaction, and improvisational structure, the paper examines how agency is distributed across performers, programmers, neural networks, training data, and technological infrastructure in live contexts. It argues that these performances do not demonstrate autonomous machine creativity; rather, they reveal creativity as a relational, materially embedded process in which the voice operates as a privileged bridge between biology and technology. Ultimately, the performances are read as a way of dwelling in the extimacy that constitutes voice and subjectivity alike, offering musicology new conceptual tools for understanding AI-mediated composition and performance as distributed yet asymmetrical co-creation.