This presentation uses your webcam and browser-based hand tracking to control slides with gestures. Camera processing runs locally in the browser. No video data leaves your device.
Move your left hand to the left to go back. Move your right hand to the right to advance. You can pause gesture controls at any time.
Designing Embodied AI Interfaces
The interface is simple on purpose: one hand, one direction, one intention.
When we call AI a tool, we imagine control.
When we call it a machine, we imagine operation.
What if it is neither — but an environment?
Every interactive system needs a grammar: what enters, what transforms, and what responds.
Gesture, touch, sensor, voice, gaze, distance, image, pressure, rhythm.
Rules, mappings, code, machine learning, interpretation, thresholds.
Light, sound, text, movement, visual feedback, robot action, spatial response.
In both paths, the body becomes part of the interface — not only the user of the interface.
Electronics make interaction tangible: the interface has weight, friction, wires, delay, resistance and surprise.
With a browser, a webcam and a few tracked points, designers can prototype embodied interactions very quickly.
Pinch distance becomes speed.
Open the distance between thumb and index.
← moves left
Open the distance between thumb and index.
moves right →
Some gesture systems already exist culturally. In China, numbers from 1 to 10 can be shown with one hand — instead of inventing gestures from scratch, we can learn from existing embodied symbolic systems.
Move an open hand in the center to rotate the cube.
A face filter about the aesthetics of power.
The hand does not only control the screen. It can control a body in space.
At this point, the question becomes less technical and more political:
What does it mean to operate a human body as an interface?
Brazilian artist, data scientist and creative technologist based in Switzerland.
What interaction would you create if the body — not the keyboard — was the prompt?