I once tried to explain Git to a ballet company. That went about as well as you’d imagine. For years, I kept my two worlds in separate rooms: in one, the visceral, human chaos of choreography; in the other, the clean, predictable logic of code.
Then, at 3 AM in my studio, staring at motion capture data flowing across a screen, the wall between those rooms dissolved. An AI system was trying to predict the next eight counts of a contemporary dance phrase, and it was... surprisingly good. Not perfect, but provocative. It was suggesting pathways my own creative habits would have never considered.
That’s when I realized the most interesting question wasn’t "Can a machine create art?" It was, "Can a machine help me make art I couldn't have made on my own?"
The Myth of Replacement vs. The Reality of Collaboration
Let's get this out of the way. When I talk about AI in choreography, people often jump to a dystopian vision of robotic dancers replacing human artists. I call bullshit.
You know what actually corrupts art? The desperation of not knowing how you’ll pay rent. The fear that makes you say yes to uninspiring gigs. An AI can’t feel the floor beneath its feet or the weight of a story in its bones. But it can be an incredible partner in the creative process, a collaborator that helps us become better architects of our own work.
So, let's strip this down to its essence. How does this partnership actually work?
Think of it as a new kind of improvisation.
- We Teach the Machine the Steps: Using motion capture, we feed the AI hundreds of hours of movement—from traditional forms like flamenco to contemporary improvisation. It learns the "grammar" of a style, the same way you learn the fundamentals of ballet at the barre.
- The AI Riffs on the Theme: Once it understands the rules, we ask it to generate new phrases. It’s like a musician improvising over a chord progression. It suggests variations, extensions, and fusions that exist outside our muscle memory.
- We Curate and Choreograph: This is where the magic happens. The AI might generate 100 ideas. 90 of them might be mechanical nonsense. But the other 10? They're sparks. A bizarre transition that unlocks a whole new section. A rhythmic pattern that feels both familiar and alien. Our job as artists is to find those sparks and breathe human intention into them.
From Flamenco to a New Language
In a recent project, I trained a model on traditional flamenco footwork—the intricate rhythms, the sharp accents, the proud architecture of the form. I then asked it to create a series of new zapateado sequences.
What it produced was fascinating. Some of it was junk. But nestled in the data were these strange, beautiful hybrids. They followed the rhythmic laws of flamenco but twisted them into something new. One sequence, full of impossible-sounding syncopations, became the centerpiece of the final performance.
When the dancers learned it, they described a strange feeling: like discovering a lost dialect of a language they’d spoken their whole lives. That’s the "click of alignment"—when technology doesn't just serve the art but expands its vocabulary.
You Don't Need a PhD, Just Curiosity
I know, I know—AI, machine learning, motion capture. It all sounds intimidatingly complex. But here’s the secret: the barrier to entry is lower than ever. You don’t need a deep-tech background to start exploring.
Projects like Stanford’s EDGE Dance Animator are creating tools that can generate physically plausible dance sequences from any piece of music, while organizations like the MIT Media Lab are exploring "human-AI co-dancing" to evolve cultural heritage. The spirit of this work isn't about replacing choreographers but empowering them with new tools for sketching ideas.
Here's my ridiculously practical, non-intimidating toolkit for getting started:
- For Motion Tracking: Your phone camera and open-source tools like MediaPipe.
- For Building Models: The most important tool is Python, but the real workhorse is your own curiosity.
- For Real-Time Visuals: TouchDesigner is the industry standard for a reason. It’s a playground for interactive art.
The point isn't to become a data scientist. It's to remain an artist who is willing to experiment. Your first ten attempts will probably fail. That’s not a bug; it’s the feature. That’s how you learn.
From Performer to Architect
Here's the philosophical shift that changes everything: moving from seeing technology as a simple tool to embracing it as a collaborator.
A hammer is a tool. It does what you tell it. An AI collaborator, on the other hand, talks back. It surprises you. It pushes you out of your comfortable creative ruts. It has endless energy for the question, "What if we tried this?"
This transforms your role from just a performer of steps to an architect of creative systems. You're not just making a dance; you're designing the process that makes the dance.
At my agency, CENIE, this is the core of our work. We build frameworks that allow this kind of human-machine collaboration to flourish, making it accessible to artists who don’t have engineering backgrounds. Because the future of creativity shouldn't belong only to those who can code.
So here's the only question that matters: what are you going to build?
Go make it happen.