Collaborating with Machines: Hybrid Performances Allow a Different Perspective on Generative Art
Additional Document Info
In the simplest view of generative art, an artist creates software, enabling it to make some aesthetic decisions on its own, and then sets it in motion, allowing the software to take over and generate the work we end up seeing. A purist might consider it cheating for the artist to intervene at any later stage. As an improviser and frequent participant in interdisciplinary collaborations with humans, I find the same thrill in collaboration with humans as I find when developing generative agents: the pleasant surprises of unexpected results I never would have thought of on my own. This has led me to explore a variety of possible collaborative relationships with my generative agents, and this has allowed me to reframe the simple view of generative art-making as one point on this spectrum of collaborative relationships. Whereas in traditional art, the human artist is fully responsible for the pre-production and the final presentation of the work, the simplest view of generative art has the human artist responsible for pre-production and a machine (created by the artist) fully responsible for the final presentation. In actuality, the pre-production phase most often involves a feedback loop in which the artist constructs a system, sets it in motion, evaluates its output, and adjusts the system to behave differently. Some artists, like composer Brian Ferneyhough , have left the machine in the pre-production stage, using it only to produce raw material, and manually shaping it to create the final presentation. In contrast to the above models, works like John Cages Inlets (Improvisation II) and Michel Waisviszs instrument the Kraakdoos (cracklebox) have indeterminate (machine-dependant) decisions built into every performance, leaving the human performer to wrestle with the machine to mutually arrive at aesthetically pleasing results during the performance. Examination of some of my compositions elucidates and expands this spectrum of possible relationships between human and machine in the final presentation of the work. These performances involve techniques such as complex feedback systems, live coding, physically and mentally strenuous performance conditions, and the grain of natural and social phenomena. In this framework, the traditional generative art model can be seen as one point in a rich spectrum of human-machine relationships, and this can pave the way for exploring new relationships. This also allows for reflections on the aesthetics of generative art: the authenticity of the machines product and the effort of the artist and performer.