How Instrument Ontologies Shape the Embodied Learning of Generative AI in Creative Work

Abstract

AI shifts musical creation by changing how performers and composers engage with instruments. Interviews with nine composers show three ways AI becomes embodied knowledge, moving resistance from physical effort to probabilistic navigation and reshaping affect and understanding.Composing and playing acoustic instruments demands embodied and cognitive investment; creating music with generative AI more often involves prompting, steering, and iterating outputs. Making with GenAI seldom feels like composing at an acoustic piano. Drawing on nine in-depth interviews with artistic composers across classical, experimental, and digital practices, we identify three instrument onto-epistemologies that shape how AI is understood, embodied, and learned. Interviewing both AI users and non-users, we trace how generative systems enter (or are rejected in) situated creative work.First, a resistance-anchored onto-epistemology treats the acoustic instrument as a partner that “pushes back.” Skill emerges through bodily struggle, repetition, and material constraint; GenAI can feel frictionless, even irritating, because it lacks weight, resistance, and energetic reciprocity. Second, reconfiguration welcomes AI’s rearrangement of musicianship. Instead of motor mastery grounded in tactile feedback, practice becomes probabilistic navigation: parameter tuning, exploratory search, and curatorial selection amid algorithmic abundance. The body does not disappear; it relocates from physical calibration to attentiveness and evaluative listening. Third, integration, most common among our respondents, reorients embodiment from resisting materials to engaging outputs, where aesthetic judgment, pattern recognition, and iterative curation become central sites of skilful participation.We suggest that GenAI can reorganize what counts as “resistance” in musical work and, with it, the forms of embodied attention and affect that structure playing and composing. These onto-epistemologies generate distinct encounters with AI, such as irritation, playfulness, or ambivalence and broadly frame creativity as embodied negotiation of constraints and possibilities rather than abstract algorithmic procedures.