“Accumulative Collaboration - Intuiting Hand Choreographies” Using machine knowing to design for contagious embodiment.
“Accumulative Collaboration” is both experimental and site-specific research in which we investigate “embodied knowing” alongside “machine knowing”.
In “Accumulative Collaboration”, we are collaborating with an open-source, neural network (a form of artificial intelligence) to facilitate human to human connections and embodied knowing. Participants are asked to improvise gestures with their hands in front of a computer to train a neural network, which learns the movements.
The hands train the neural network and a duet between machine and human ensues.
The hand improvisations become training data for the neural network. In turn, the neural network facilitates an accumulative choreography -- one participant follows another, building off of previously improvised hand gestures. A contagion of choreography? We hope so!
We (you as participants and we as researchers) are training the neural network together. A Leap Motion Controller will detect hand movement and send this data to Wekinator (an open source machine learning tool).
Our experiment asks:
How do we bring ideas of “machine knowing” with “embodied knowing” into conversation?
More broadly, how do we unpack and widen discourse around A.I.?
In the context of A.I., how do we re-contextualize the physical body as an entity with its own rhetorical agency?
And perhaps our favorite.. What is a radical AI?
The barrier to “improvising” with one’s hands seems to be lower. Audience members do not need formal movement training to “dance” with their hands (compared to if we were to ask participants to “improvise” on the spot with their bodies).
On the subject of hand choreographies, we would like to acknowledge (and celebrate):
Credits (open source tools): Wekinator - Rebecca Fiebrink, Darius Marowiec - (NOK) & Processing