Audience As Virus

Audience As Virus is an interactive immersive installation-performance that involves the audience in the possibility to break AI predictions of human movement and offers the public a clear capacity to intervene. The work misuses ChatGPT and HMDM (Human Movement Diffusion Model) to counter the patterns of replication, mimicry, and averaging. Drawing on practices often excluded from the Art & Technology canon—from Capoeira, Candomblé, and circus traditions in Brazil to experimental choreography in the USA—to “counter-predict” dominant models and question which bodies and gestures are made invisible when AI decides what counts. Embracing complexity, complication and excess, this participatory work uses anti-efficiency and errors to provoke new unpredictable, perhaps absurd, (but imaginable) movement and expressive relationships. A performer initiates a gesture. The AI attempts to continue the movement, to mimic. The audience intervenes with movement, sound, or type. Each presence becomes a trigger for an unhinged mix of prompts leading to a frictional situation where movement is expanded, layered, unexpected, and affective, refusing to resolve. Audience As Virus is built on a custom multilayer AI pipeline using ChatGPT and HMDM, poetic prompts, and a database of choreography and dance texts. It invites the public to break, interrupt, and reconfigure AI predictions of movement in real time. What happens when gesture becomes multitudinal? When group relations are formed not through clarity but through complication? In Audience As Virus the paradox of counter-prediction becomes method. In the same way utopias are impossible but absolutely necessary, the intent for (the impossible) counter-prediction is a continuous act of reclaiming free action. In a Deleuzian twist, the project frames the space as a field of potential—where bodies don’t express identity, but produce difference. Suumary: Audience As Virus is an interactive immersive installation-performance that involves the audience in the possibility to break AI predictions of human movement and offers the public a clear capacity to intervene. It misuses ChatGPT and HMDM (Human Movement Diffusion Model) to counter the patterns of replication, mimicry, and averaging. Collaborators: Danielle McPhatter (EY Intelligent Realities Lab): Creative technologist Vinson Fraley: Choreographer, performer Lucie Strecker & Angewandte students: Choreographer, performers at Vienna venue Dafna Naphtali: Sound Selwa Abd: Sound Wobbly: Sound Matthew Gantt: Sound Daveed Baptiste: Physical costume MX Oops: Digital costumes Deandra de Anthony: Lighting at CultureHub NYC and ISSUE Project Room Mattie Barber-Bockelman: Master of ceremony at CultureHub NYC and ISSUE Project Room The collaboration continues