Entoptic Field Monitor

Utilizing Machine Learning Uncertainty as a Design Material

This project is now archived and was part of DDW22

This project is part of
Design United | EXPO

Entoptic Field Monitor (screenshot)

The Entoptic Field Monitor (a post-GDPR surveillance application) is a research product utilizing AI technology. It is the result of a metaphor-driven design process, which seeks to make explicit how AI technologies shape particular realities---how these are perceived and made sense of.

Probing Realities shaped by AI Technologies through Design

AI technologies are extremely difficult to conceptualize and apply to both design methods and critical observation. People working in technical fields usually tend to focus on AI technology in and of itself to explain or make it transparent. However, the more subtle, everyday ways in which AI technologies shape human experience (such as in smartphone image improvements) often go unnoticed. In this project, I address this by extrapolating these ‘everyday’ occurrences, creating prototypes that provoke reflections on what AI technologies do to change our view of the world. Grounded in the material analogy of entoptic phenomena, we developed a prototype which takes incoming video images and renders synthetic outputs using a Generative Adversarial Network. It is a low latency camera application that monitors a space and renders AI-based approximations of the camera feed. The application runs on a tablet that faces into the space, and can be mounted on any suitable surface. By neither storing nor displaying any of the incoming data, the Entoptic Field Monitor raises questions around privacy, representation and acceptability; as well as what it means to design with AI technologies.