When
Thematic Session 1: Design and Interaction (Monday, 14:15)
Abstract
A holistic approach to AI, in this context, is one that considers all processes in machine learning as creatively important to the musician. This encompasses model architecture design, data curation, hardware curation, training, inference, the coupling between model and external environment, and discovery of potential tuning points in a model. This perspective strays from a more conventional approach where engineers design models for musicians to use. Instead, one might find creative opportunities in all AI processes up to and including inference. In terms of user interaction, the interfaces to broader broader processes are typically spread across multiple components, across code libraries and text editors for model and data curation, across physical interfaces for data collection and inference. These interfaces may be disconnected in a way that might preclude a good quality of embodied interaction. We can question how important it might be that we can engage with this collection of processes in an embodied way. What are the best ways to open up these processes together for creative work? For example perhaps it's valuable to unify training and inference into a single instrument, such they can both be approached with the same level of musicality. Perhaps we should look at mid-level interfaces that provide unified, embodied interactivity with data curation, training and inference? What technologies can enable us to embed these processes in physical devices? This presentation will explore these questions through examples from musical practice.
Bio
Chris Kiefer is a computer-musician, musical instrument designer and Lecturer in Music Technology at the University of Sussex. As a live coder he performs under the name "Luuma." He plays an augmented self-resonating cello as half of improv-duo Feeback Cell, and with the feedback-drone-quartet "Brain Dead Ensemble."