Mirrored Intelligence or Intelligent Mirror?: Historical Perspectives of AI-Based Agents in Multimedia Performance

The present article discusses the application of AI in multimedia performance from a historical perspective, analysing the use of intelligent algorithms since the 1940s in artworks that implies computer programs designed to be functionally autonomous. It addresses the evolving notion of AI according to sociocultural processes and technical implementations and focuses on how it became observable in certain features of live plays. The authorship behind these artworks shows how the concept of ‘intelligence’ applied to computer programs has changed, entailing a different embodiment of the algorithm – i.e. its manifestation through staged devices – and human-machine relationship – i.e. the interaction with author, performer and audience before or during the play. The paper highlights some salient phases of this process as occurring in Western culture. The starting section argues that AI was first conceived as a stand-alone device and the relationship with the machine was mainly the prerogative of the authors. It will be then underlined how, due to the interest in the extemporaneous representation extended to generative algorithms, a dynamic interaction between AI and other agents has been progressively entailed. The last part will concern the technical development of machine learning and its increasing accessibility as crucial for addressing today notion of artificial intelligence. The relationship with AI agents in most recent computer-generated works has also raised ethical aspects, both when the algorithm is explicitly enacted and when it is used in the background, as equally imbued with socio-technical implications related to hybrid ecosystems.