Generative Music as a Catalyst for Social Choreography

Z’s MIT thesis explored how generative AI can support new forms of social interaction through music and dance. The project introduced three motion-to-music systems: Con Moto, a real-time embodied music system for duet performance; MoveGPT, a causal transformer model that generates music from live depth-camera motion; and Encypher, a group cypher system that maps collective movement qualities into prompts for live music generation. The work was evaluated through autobiographical design, five weeks of co-design with local dancers, an IRB-approved user study, and a public museum event. Across the thesis, Z foregrounded sociality, co-design, and embodied interaction as central concerns for building AI systems that support collaborative creative expression.