Self Portrait by Ben Cullen Williams London Design Festival
Self-Portrait is a 36 minute video artwork by London based artist Ben Cullen Williams, created in collaboration with Google DeepMind Research Scientist Jason Baldridge. The work presents a self-portrait of generative AI by using Google’s generative AI tools, fine-tuned and informed by Williams’s archive of his own photographs.
3 minute excerpt of ‘Self-Portrait’
Self Portrait by Ben Cullen Williams London Design Festival
The 36 minute video loop was first presented at Design Culture, as part of London Design Festival 2025 on a 7m x 2m LED screen.
Self Portrait by Ben Cullen Williams London Design Festival
The project delves into the centuries-old question of what a self-portrait is and what it can be, using the capabilities of generative artificial intelligence to explore themes of landscape, identity, technology, human endeavour and expression.
Self-Portrait with Striped Shirt (1910) by Egon Schiele Leopold Museum
It builds upon the history of self-portraiture, from Jan Van Eyck's 15th-century figurative painting, "Portrait of a Man", to Egon Schiele’s expressionistic fluid paintings “Self-Portrait with Striped Shirt”, and Louise Bourgeois' abstract 20th-century sculpture, Torso, "Self-Portrait", which represents a personal sense of self rather than anatomical precision.
Self Portrait by Ben Cullen Williams London Design Festival
Discussions between Williams and Baldridge began in 2021, seeing if a model at the time, ‘Parti’ could create images of the network that it was formed of. Simple prompts were used to see what visual style the model outputted with minimal human suggestion.
Self Portrait by Ben Cullen Williams London Design Festival
In early 2025, Williams and Baldridge picked up discussions about these 2021 ideas. They started to explore which of the latest models could help bring these ideas to a point of actual realisation.
The process began with a series of in-depth interactions between Williams and Google’s AI, Gemini. Williams prompted the tool to describe itself, leading to a dialogue that spanned a spectrum of subjects—from the physical hardware and infrastructure that constitute it, to the people who build and maintain it, and from immaterial signals such as light and radio, to societal phenomena amongst many others.
The information given by Gemini about generative AI was then converted into detailed image prompts, which were used to create the visual framework for the project.
3 minute excerpt of ‘Self-Portrait’
To define the visual world a series of models were fine-tuned using Williams’s own photographs in collaboration with Baldridge. This process allowed Google’s generative AI image model, Imagen, to generate images that were imbued with some of the qualities of Williams’s photographs, blurring the lines between the creator and the tool.
Self Portrait by Ben Cullen Williams London Design Festival
The project intentionally raises the question: “Is this a self-portrait of the generative AI or in fact a self-portrait of Williams himself?” This duality is at the heart of the work, with the artist and the AI becoming co-creators in the exploration of self. The work doesn’t contain a linear narrative – instead it is a looping cyclical collection of archival fragments, that have been ordered by Williams himself.
This process of human, machine, human allowed Williams to question the role of the machine in the creation of the work. The subject of the work is AI, the medium is AI, is it a self-portrait of the AI or in fact a self-portrait of Williams’ own artistic practice.
Still from Ben Cullen Williams' "Self Portrait" London Design Festival
Much of Williams’ artistic practice is preoccupied with how technology, both analogue and digital, shapes how we perceive and capture the world. His work often alters and disrupts the material processes of technology. Williams saw the fine-tuning of the models as an experimental process akin to working in a photographic darkroom.
Ben Cullen Williams Self-Portrait Still 10 London Design Festival
The AI model was treated as a raw, malleable material that could be sculpted and transformed to create unexpected outcomes. The longer the model was exposed to his images, the stronger the impression. The AI model is seen as another step in the technological processes within the artist's image making. From medium format film, and chemical baths in the darkroom, to digital scanners and now through algorithms.
Self Portrait by Ben Cullen Williams London Design Festival
As a result, a profound aesthetic shift was created, producing a work that contains images from across a wide spectrum—from affected to unaffected—constantly oscillating between the visual language of the models and that of Williams.
From the hundreds of generated images, Williams carefully selected images that were then fed into Flow, Google’s AI filmmaking tool, powered by Veo to bring them to life. Williams chose images that he felt had a physicality to them, something he felt connected back to the photographs that were part of the training set.
Self Portrait by Ben Cullen Williams London Design Festival
Within Flow, Williams directed the shots with specific cinematic instructions in order to create the movement and atmosphere he needed for each shot.
Self Portrait by Ben Cullen Williams London Design Festival
The wide screen format of the video reflects this connection to cinema. The duality at the heart of this video work is reflected in the inclusion of the cinematic split screen.
3 minute excerpt of ‘Self-Portrait’
Self Portrait by Ben Cullen Williams London Design Festival
Williams worked with two musicians Harrison Cargill and Gaika Tavares to develop the score for the work. Created to accompany the AI generated foley sound textures, the music was intended to emphasise the raw sensibility of some of the shots while giving space for the visual and textural world to come to be at the foreground.
You are all set!
Your first Culture Weekly will arrive this week.