The Material Gaze
Reflective AI as a Design Practice
As a designer, I came to artificial intelligence out of fascination: I wanted to understand how machines see, and what that might reveal about how we, as makers, do. But today in creative practice, AI is mostly used not to look closer, but to go faster, or even to automate the creative process altogether. In this thesis, I develop an alternative approach: a practice called Reflective AI, which slows the technology down and uses it not as a tool for production, but as a site for reflection. It treats often-overlooked stages of AI practice (from collecting data to training a model) as a material to be worked, the way a ceramicist works clay.
This approach is developed across five projects that range from investigating object detection models to a three-month ceramic residency. Throughout, I use machine learning as a mirror: I train models on my own subjective judgments, and then study how the models reflect my ways of seeing back to me. What emerges is a material gaze: a way of seeing that comes from the making of and with the model, and looks back on the maker. This thesis offers a way past the question of whether AI is good or bad for creativity, toward a slower, more considered relationship with it.
