The research work presented in this paper is at an early stage. While tests for artificial intelligence/consciousness/sentience continue to evolve, there still doesn't seem to be any consensus on what exactly we hope to capture with such tests. In this paper, we explore the possibility of an arts-based performative test for machine sentience. We analyzed a specific choreography, called Chemins à coulisses, through both the language and experience of the choreographer and a spectator, and we offer some additional theoretical lenses through which to tackle this problem, including phenomenology and distributed systems. We offer the beginnings of an approach to creating a performance-based test where the machine doesn't merely imitate steps but engages in a meaningful way with performers and spectators. There may be an under explored area for benchmarking machine sentience that sits at the intersection of phenomenology, choreography, and distributed systems.

No abstract