When Movement Meets Machine
What happens when improvisation, a deeply human, spontaneous act is extended into the realm of artificial intelligence? Could a machine ever become a meaningful creative partner?
This project wasn’t about solving a problem. It was about creating a new kind of collaborator, one that pushed boundaries, challenged assumptions, and expanded the language of movement itself.
How might we embed AI into embodied creative practice without automating artistry or diluting expression?
This project positions co-creative AI not as a performance gimmick, but as a strategic catalyst for creative reflection and movement innovation.
TLDR
This project reimagines how dancers can engage with artificial intelligence as a creative partner, embedding LuminAI into a university improvisation class over 7 months to study its impact on movement, expression, and co-creation.
​​
The solution
An AI system integrated into dance pedagogy that enhances creative exploration, supports self-awareness, and redefines how technology can augment artistry.
​
My Role
Led the end-to-end redesign and field study: from user research and interface design to diary study analysis and final performance.
​
The Context​
​Georgia Tech & Kennesaw State University, funded by the NSF; interdisciplinary collaboration across HCI, dance, and cognitive science.
Want a more detailed run down of this project?
A Designer’s Challenge
As a strategic designer, I wasn’t just building an interface, I was crafting a relationship. I needed to understand what dancers value in improvisation, identify friction points in human-computer movement collaboration, and evolve a system that could both provoke and support their creativity. This meant not just designing the tool, but embedding it within a curriculum, facilitating user reflection, and building long-term engagement with the AI.
​
This project culminated in the world’s first collaborative, improvised dance performance between humans and AI.
Understanding the Dance-AI Gap
We began with a foundational study involving 5 experienced dancers and a choreographer, observing how LuminAI fit into their improvisational routines.
-
Pain points: lack of visual grounding, jarring tempo changes, limited personalization, and unclear AI feedback.
-
Insight: Dancers didn’t want a machine that simply mimicked, they wanted a partner that inspired, challenged, and reflected their unique styles.
​
"It made me think of Just Dance... how each mode had a motive like high energy or slower movements” - D1
"If I’m working in a particular area a lot, does it start to turn a different color, like a mood ring?” - I
​
To dive deeper, we embedded LuminAI in a semester-long university dance course. Through 13 sessions and over 240 reflection entries, we asked dancers to not only move but also reflect. This diary study helped us map emotional journeys, usage patterns, and creative transformations.
Preliminary Study: Introducing LuminAI to dancers
Redesigning Co-Creation
We redesigned LuminAI across four strategic axes:
-
Visual Grounding: Introduced a “virtual floor” and energy-based visualizations that made AI movements easier to follow and feel more embodied.
-
Tempo Transparency: Added speed controls and visual feedback to help dancers understand how the AI was processing their movements and align more intuitively.
-
Personalization: Created a tagging and memory system so LuminAI could adapt to each dancer’s style and log past movements, laying the groundwork for individual AI avatars.
-
Spatial Awareness: Prototyped color-based cues that mapped spatial interaction patterns, a way for dancers to "see" how they were moving through the room over time.
Design Iterations
Implementation in the Classroom
We didn’t just study how dancers used LuminAI, we used that research to shape its evolution in real time. Over 13 class sessions, we embedded LuminAI into a structured improvisation curriculum and designed a longitudinal diary study to observe how creative behavior unfolded across time, space, and reflection.
This wasn’t about usability testing, it was about co-developing the tool with its users. We intentionally chose a diary study format because it allowed us to go beyond first impressions and track dancers’ evolving relationships with LuminAI, mapping their shifts in mindset, movement, and creative agency.
​
Each 75-minute session was structured around:
-
A warm-up and conceptual introduction (e.g. spatial tension, energy, or compositional tools)
-
Improvisational exercises with and without the AI
-
Guided interaction with LuminAI (solo, duet, and group)
-
End-of-class journaling using emotion wheels and open-ended prompts
These reflection logs, alongside researcher observations and mid-semester interviews, helped us surface strategic insights.
​
We discovered:
-
Students began embracing simplicity over complexity.
-
They used LuminAI to break out of habitual patterns.
-
They developed deeper body awareness and spatial understanding.
-
They began to see the AI as a creative partner, not just a tool.
Dancers engaged with multiple creative sub-processes: generation, elaboration, association, evaluation, and creative metacognition, often simultaneously. These were not just moments of performance, but moments of strategic reflection and adaptation.
​
“I saw its usefulness as a tool and changed my perspective from ’limited’ to ’simple.’ Sometimes we overcomplicate things as dancers when it comes to improvisation. During the semester, I felt that the main theme was ’going back to basics.’ Rooting myself in the movement - AI helped me accomplish that." - D4
​
"It influenced the movement generation. I really enjoyed the exercise where we embodied the avatar’s movement, learning what it had created. This challenged our timing and capabilities, as the avatar often makes choices that are jerky or static, and it also creates impossible shapes with its body – it can be interesting and fun to try to embody the impossible." - I
Final Showcase
The semester culminated in the world’s first collaborative, improvised public performance between dancers and a co-creative AI. Dancers had developed personal strategies for engaging with LuminAI, often treating it like a duet partner, mimicking its “impossible” shapes, interpreting its unpredictability, and incorporating its rhythm into their own compositions.
​
This wasn’t just a demo of tech. It was a story of trust-building, of co-authorship, of reimagining how we learn to move and how we move to learn.
Future Directions
LuminAI shifted how dancers perceived both AI and their own creativity. Participants reported increased self-awareness, a deeper engagement with foundational techniques, and a willingness to break habitual movement patterns. More than a tool, LuminAI became a collaborator, one that helped dancers slow down, reflect, and experiment. The final performance, the first of its kind, showcased improvised duets between humans and AI, signaling a new frontier in performative arts.
Looking ahead, this project lays the groundwork for scalable co-creative systems in education and performance. Future iterations will explore personalized AI avatars trained on individual movement histories, multi-user improvisation, and delayed feedback mechanisms to support reflective choreography. As AI continues to move from novelty to necessity in creative domains, projects like LuminAI ask: how can technology not just augment artistry, but become a part of its evolution?
​
Reflections
LuminAI wasn’t born to solve a problem, it was created to pose new questions. What if improvisation didn’t stop at the edge of the body? What if an AI could not only observe, but participate, influence, and respond?
Over the semester, dancers began to treat LuminAI not as a passive tool, but as a living presence in the studio, one that nudged them toward discomfort, sparked curiosity, and made them see their own movement vocabulary with new eyes. It wasn’t about replacing human connection, but expanding its possibilities. LuminAI became a mirror that moved, a duet partner with no ego, a provocation in digital form.
This project didn’t just reimagine AI in dance, it reframed what co-creation could look like in the age of computation. And as dancers adapted, reflected, and improvised in response, they helped design a future where movement and machine co-evolve fluidly, creatively, and unexpectedly.