Self Portraits in AI

This project began as a fairly practical problem: I needed a headshot. Or, more precisely, I was in a moment of my life where I felt newly pressured to become visible in a particular way. I had recently finished my PhD. I was navigating academic job searches, speaker bios, websites, social media profiles, and all the small professional rituals through which a person becomes publicly legible.

At the same time, I was emerging from a serious illness, and my relationship to my own face, body, and professional future felt unstable. So the project began with a question that was both vain and existential: what image of myself was I supposed to send out into the world?

I am not interested in AI as a mirror, because the mirror was already the wrong metaphor. The self-portrait is a better one: not a reflection of a stable self, but a constructed encounter with subjectivity, framing, materiality, and difference.

In AI self-portraiture, hallucination is not simply a mistake layered onto an otherwise faithful image. It is the medium. The face came back as a probabilistic fiction: recognizable enough to be mine, strange enough to show me that likeness is not the same thing as truth.

This project unfolds over the course of a year, employing a research-creation methodology that integrates computational, analogue, and public practices. It begins with self-photography and generative image work across multiple AI models, recording parameters, and archiving proliferating outputs. These portraits are then subjected to both photoshop remixing and analogue redrawing, where sketching, erasure, and reworking reintroduce embodied friction and human trace.

Alongside these practices, I have kept autoethnographic notes—process diaries, refusal logs, and reflections on ethical dilemmas—that document the tensions of working with one’s own image as data. Crucially, the project has been staged publicly throughout its evolution: blog posts and an early conference presentation have served as provisional pedagogical platforms, enabling dialogue and critique.

The primary data are my own genAI mediated images, sketches, and process documents, including prompt logs, refusal notes, and expressive writing. These materials are treated as both aesthetic artifacts and analytic traces of entanglement.

The project’s purpose is threefold: (1) to create a public account of method and process that makes visible the entanglements of machine, body, and imagination; (2) to offer educators a model for engaging AI not as a technical problem but as a cultural and aesthetic site for reflection; and (3) to challenge prescriptive discourse by performing vulnerability and playfulness as experimentation.

By situating the shifting self as data, the work insists that creative practice is also critical practice, capable of holding both anxiety and possibility in our encounters with generative AI.