Prompt Dancer * 2023
Theater on Orlí, JAMU Theatre Faculty (Brno, Czech Republic)
Prompt Dancer is a live, experimental performance and research showcase in which a contemporary dance choreography is generated entirely by an AI system and performed without prior rehearsal. Created by António Baía Reis, Sára Solmošiová, and Petr Vašků, the project stages a deliberately asymmetrical relationship: the AI occupies the role of choreographer and director, while the dancer becomes an interpreter under pressure, receiving step-by-step instructions through a synthetic, robotic voice and translating them into embodied movement in real time. The work is designed to expose the fragile, unpredictable space where algorithmic language meets somatic intelligence, and where “instruction” becomes performance only through the dancer’s interpretive labour.
Rather than reproducing media narratives about AI either “replacing artists” or “saving creativity,” Prompt Dancer moves the debate onto the stage as lived experience. Dance, an art form grounded in sensation, timing, breath, and tacit knowledge, is placed in productive friction with a language-based machine that outputs text. The dancer cannot ask for clarification, negotiate, or refine the choreographic material through dialogue; they must immediately comply, resist, improvise, or adapt. This constraint foregrounds the project’s central tensions: authority vs. embodied knowledge, command vs. interpretation, efficiency vs. presence, and algorithmic coherence vs. human meaning-making. Each iteration is intentionally unstable: prompts mutate, tools evolve, and the dancer’s responses remain contingent and unrepeatable, making the piece both performance and “living experiment.”
The project’s development began with an artistic residency led by António Baía Reis at JAMU Theatre Faculty in Brno, Czech Republic (October 2023), where interdisciplinary participants explored AI’s relationship with creative practice. The team then analysed the residency through arts-based methods, narrative diaries, experimental video, and photography, to distil emergent themes. These themes became the material fed into ChatGPT to generate choreography. The instructions were recorded with a synthetic voice and pilot-tested in a dance-class context, before the official presentation at the JAMU Conference 2023, which combined the live performance (15–20 minutes) with a methodological presentation and a public discussion.
As an arts-based research project, Prompt Dancer also produces reflective knowledge: it documents the methodology of AI-choreography generation; collects and analyses the experiences of dancer, audience, and creators; and examines what this setup reveals about creative agency, labour, obedience, and the co-production of meaning in AI-mediated performance. In its simplest formulation, the work asks a looping question, humans prompt machines, machines prompt humans, and humans create what?, and answers it not with theory first, but with the immediate evidence of bodies, voices, and attention sharing time in the same room.
MEDIA: