PROMPTLESS
Physical Interfaces for Human–AI Visual Co-Creation
What if you could touch AI instead of prompting it?
PROMPTLESS is an interactive media-art project in which knobs, buttons, switches and sensors shape an evolving visual system. Participants use touch, rhythm and continuous adjustment to influence what they see. The work explores the relationship between human intention, computational behaviour and AI-assisted visual material.
A physical action.
An immediate visual response.
Each movement changes a visible parameter: spatial arrangement, density, speed or distortion. Participants can explore the system through repeated gestures and observe how their decisions transform the visual field.
The proposed physical controller connects to a browser-based generative engine. Optional light or ambient-sound sensors can also bring the surrounding environment into the work.
AI-assisted material forms a separate visual layer. Its updates are handled independently of the continuous rendering, so model processing does not interrupt the immediate response to physical input.
Touch
Turn knobs, press buttons or change a switch on the physical controller.
Translate
An ESP32 or Arduino sends input over USB serial, mapping each gesture to a visual parameter.
Respond
The browser engine continuously changes the composition, movement and visual behaviour.
Experience
A screen or projection makes the response visible. The participant adjusts again, shaping the next state.
Small adjustments.
Different possibilities.
The controller is intended to make the connection between action and outcome easy to explore. Its precise mappings and control ranges will be refined through prototype testing.
The research asks how embodied interaction changes a participant’s sense of agency, authorship and understanding of an AI-assisted creative process.
| Composition | Spatial arrangement |
|---|---|
| Complexity | Density and visual detail |
| Movement | Speed and rhythm |
| Distortion | Deformation and variation |
| Human / AI | Balance with AI-assisted material |
The audience becomes
part of the research.
The project is designed to develop through open studio sessions, hands-on exchanges and public testing. Participants can try the interface and discuss what feels intuitive, surprising or unclear.
These exchanges will help refine the relationship between physical gesture and visual response, while opening a conversation about how authorship is shared between a person and an AI-assisted system.
Concept developed. Physical prototype in development.
The first working physical controller has not yet been built. The next phase connects the hardware, generative visuals and AI-assisted material, then tests the experience with participants.
The live website is a procedural browser study of the visual response. Visitors can adjust composition, density, movement and distortion, scroll into the particle field and capture a visual state. Hardware and AI-assisted image integration are planned for the physical prototype.