Work > Publications > A.I. charmer
A.I. charmer
A.I. charmer is a performance and an exploration of the relationship between body behaviour and artificial intelligence.
This experiment's necessity came from the oddness I feel every time I look at the documentation of my performance Migrant Market and the observations I received about my video performance Automated pain?.
What all these performances have in common is the tool I used, an AI-powered gimbal camera, an inanimate object using mainly face tracking.
During the performance Migrant Market, I activated the tracker by accident. The consequence was an obvious tension between the filmmaker who was recording her point of view, physically engaged in the space with subjectivity and sensitivity, and the A.I. trying to analyse the right course of actions.
The A.I. followed a series of protocols based on light, the field of vision, and the number of objects and living creatures in the space. The result shows a succession of motion and intention counter orders. None of the decisions are wrong, but there is confusion between who/what the subject and the object are.
Automated pain? is a video performance about digital design and premeditated emotions. The role of the gimbal camera was to be a witness and an archivist. The camera was following my first expression of freedom after a week without any digital device.
As I was presenting my work at the Research Academy Performative Practice (The Situational Self - Acting and Identity) organised by the Zurich University of the Arts in late 2020, Hong Konger dramatist Donald Chung questioned the veracity of the document and specifically how the choice of an A.I.-based tool could take the truth away from liveness.
With A.I. charmer, I'm starting to engage with A.I., movement, and intensity.
Is a collaborative performance between A.I. and a human live or/and real?
Can the moment of the performance push the A.I. limits (without updates)?
Is empathy an adaptive factor or a form of unconscious resistance?
Is tracking making the object the de facto subject or is motion disturbing the digital framing?
What could be the audience's expectations and desire while being aware of the rules of engagement?
During my first test, I realised that the A.I. couldn't follow my two hands. One hand with multiple fingers proved to still be challenging.
I decided to use one finger and tried an invisible dance/drawing with the inanimate thinker. The gimbal camera preferred articulations based on pattern behaviours over speed and offbeat intensity. A.I. charmer questions the tool’s abilities to mimic or premeditate reality.
The link between alive and liveness is becoming more complex and confusing as the audience consumes the performance/video performance/livestream in a designed digital ecosystem establishing and policing automated behaviour via a normative interface and experience.
A.I. charmer is a work in progress exploration because A.I.-based gimbal cameras or 360 cameras are or will be a necessity for the documentation of in-house or public performances.
PS: During the next phase I will explore the ethics of stitching for 360 documentation.