Last March, I watched a contemporary piece in Brooklyn where the choreography credit read "Created in collaboration with MotionGPT." The dancer onstage moved through sequences that felt simultaneously familiar and alien—precision I'd expect from the Paris Opera Ballet, but with angular transitions no human would naturally choose. I couldn't decide if I was watching the future or a party trick.
Robots Don't Get Stage Fright
Here's what the AI choreography tools actually do: they swallow motion-capture data from thousands of performances, process biomechanics research, and spit out movement phrases. MotionGPT, DeepDance, and their competitors work in seconds what used to take choreographers weeks of sketching and studio experimentation.
Wayne McGregor's studio has been experimenting with algorithmic movement generation since 2017. His piece "Living Archive" used AI trained on decades of his own work to propose new sequences. The result wasn't a replacement—it was a conversation. Dancers rejected 80% of what the machine suggested. But that remaining 20%? Movements they'd never have discovered alone.
The Soul Question Misses the Point
Purists love to ask whether algorithms can feel heartbreak or rage. Lina Park from Seoul's AI Dance Lab frames it differently: "We're not asking the machine to feel. We're asking it to propose." The "spark" debate feels hollow when you watch a TikTok choreographer with zero dance training generate a routine that lands a brand deal. Maybe the question isn't whether AI has soul, but who gets access to choreographic tools.
The 2024 Paris Olympics opening ceremony used AI-designed synchronization for 1,000 drones and 200 dancers. I spoke with one of the performers afterward. She described the experience as "dancing with a calculator"—weirdly precise, strangely beautiful, not something she'd want every day.
Preservation Over Replacement
The technology excites me most when applied to endangered movement traditions. A team in Detroit has been archiving folk dances from disappearing immigrant communities using motion capture, then training models to generate new variations that stay true to the original vocabulary. When the last practitioner dies, something of their movement memory survives.
This isn't about replacing choreographers. It's about tools. Cameras didn't kill painting. Drum machines didn't end drumming. The dance floor will absorb this technology like it absorbed everything else—awkwardly at first, then invisibly.
I'm booking tickets for an AI-choreographed premiere in Chicago next month. I'll let you know if my skepticism survives the curtain call.
