I Resist The Body Electric: Subverting the Capture of Musical Gestures to Train AI Models
Fabio Morreale (Sony AI)*; Morgan Scheuerman (Sony AI); Giacomo Lepri (Università di Genova); Liam Pram (University of Auckland); Astrid Bin (Universität der Künste)
- The registration for this workshop is managed by the workshop chairs. A sign-up form will be circulated in late May
- Type: In person
- Room Location: 401
- Afternoon (2pm-6pm)
- Signup Deadline: June 8th AoE
- Contact Email: here
Abstract:
Generative text-to-music systems are trained on vast amounts of musical data, often extracted without consent or remuneration for musicians. While these systems can convincingly generate fixed media (recordings surrogates) they entirely lack the capacity to replicate the embodied and performative dimensions of musical practice. Live performance, therefore, remains largely outside the scope of current generative systems. However, ongoing research in robotics, combined with lurking venture capital, suggests that a future populated by surrogate robotic musicians simulating human musical gestures is far from implausible, a development that would likely require AI companies to perform large-scale collections of gestural data derived from performers’ bodies and practices in order to abstract models of gestural knowledge. This workshop explores how we might anticipate and resist the potential extraction of gestural data for training future AI systems. Drawing on anticipatory ethics and speculative design, participants will imagine hypothetical technologies designed to covertly capture musical gestures and then develop countermeasures. Through collective prototyping and discussion, the workshop aims to surface ethical, technical, and political questions surrounding gesture and data extraction in AI-driven music technologies.
Workshop website: https://resist-the-capture.surge.sh/