RESEMBLANCE is a series of portraits made with generative artificial intelligence (ai) using text prompts.

I arrived at this set of images after many years of photographing people in a conventional way, face-to-face with a film camera. I documented people who posed with signs setting out their comments on significant Australian events. As well as exhibiting the subjects’ appearance, the portraits reflected their attitudes on the theme of the day.

In this project, the event is the emergence of generative ai. I have chosen to document the inhabitants of the ai world and I have asked them to tell us about this game-changing phenomenon. The people and their messages are generated with ai but they are also the result of my discussions with ai, as well as ai’s interactions with a mind-blowing volume of human made training data. The portraits are created in the style of mid-century studio photographs and the messages are rendered by ai (it turns out ai is still perfecting its handwriting). RESEMBLANCE is a series of portraits of people created with ai, with messages from ai, about ai.

The subjects’ eerie resemblance to real people is mirrored in the resemblance of the images to real photographs. As I wondered through the uncanny valley, I came to realise the similarities were not limited to the images. The distinction between my emotional responses to the ai portraits and other real-world photographic portraits began to blur. Both gave a sense of human connection. Both inspired stories in my imagination.

I know the people are not real but I am still fascinated by how they turned out. The variations in their facial expressions, their gazes or their stances are alluring, especially when I have not specified these elements in my prompts. The way these features coalesce draws me in. They hold my attention as I scan the details in the frame in an effort to understand how they came to be. I find what emerges to be beautiful.

These synthetic images have parallels with my real-world portraits. Often, a photograph reveals more than I intended or expected. It reveals details not noticed at the time the photograph was captured — evidence of the unseen. In creating a complete picture from an imperfect prompt, images generated with ai also reveal something unexpected — they divulge evidence of the unsaid. Simple prompts don’t give simple pictures; complex prompts deliver surprises; in all cases, the details are intriguing.