Identity and Intimacy in Human-Computer Improvisation

Abstract

Artificial intelligence invites a new approach to computing in live music performance. Computers and human performers might collaborate on an equal basis. The perceived identities of participants, both human and machine, are enriched but problematic. The conflicting relationships between these identities impact upon both performers' and listeners' experience. The film Orlacs Hände is a starting point for a speculative discussion about human-computer improvisation, problems of identity, the self and the Other, social intimacy and the therapeutic process. INTRODUCTIONComputational creativity offers the prospect of proactive, artificial performers able achieve Boden's key facets of creativity -newness, surprise and value [1] -in synergy with their human partners, but not dependent upon them at the moment of realisation. Such systems must have capabilities that are not attributable to encoded rules or overt direction. A novel performance practice is possible that admits computers as equal participants, not as glorified musical instruments. This is the central aim of the Live Algorithms for Music network [2].If human-computer improvisation is genuinely collaborative, this means computers should not be tethered to controllers, no more than a human performer might expect to be. But it also poses a problem of how machines might enjoy a performer-like identity, given that this socially situated praxis is normally understood to be the exclusive preserve of humans. Arguably, there is already a crisis in established computer-based performance, a confusion of identities and a problematical relationship with audiences. We can speculate how more general models of human interrelationships might further our understanding of a "live algorithmic" performance practice.My current experiments with real-time improvisation systems began with "piano_prosthesis" (2007), the first of a series of systems for solo musician and computer [3]. These semi-improvised duo "compositions" present an environment for interaction based on mutual listening, adaptation and creative action by both parties [4]. Both participants acquire knowledge about the other's behaviour, and this knowledge is represented ("expressed") in the music they produce. For the computer, the sound world in each version is closely allied to the timbre of the relevant instrument. This is intended to be a quasi-social process. The "prosthesis" of the title is used questioningly: do we hear the electronic element as only as extension of the performer's wishes and the solo instrument's capabilities, or as another, intentional agent?

A Subsumption Agent for Collaborative Free Improvisation

“…This general approach to interaction design is shared by other systems with similar aims, such as those by Blackwell and Young (2004) and Collins (2006), although implementations vary greatly. More generally, this interaction model is opposed to "game-theory models of social interaction that emphasize self-interest", and instead emphasizes coordination, "interdependence", and "mutual control" (Young 2010). The human and computer players function as tightly coupled subsystems, exerting a constant reciprocal influence on one another.…”