AI that generates images from text is hitting the mainstream.
The artist David O'Reilly recently pondered an "infinite art scenario," a future where AI creates endless art that is personally tailored to satisfy each of us. He points out that whatever can be cheaply automated may soon enough be considered worthless:
“Consider how we tend to ignore most of the natural world and all its intricate forms, just as we do the infinite perfection of fractals or the billions of images of our planet on Google Earth, each beautiful in their own way. We seek the edges of human potential, and find meaning in what people spent time on, everything else is noise.”
I've watched text-to-image AI incrementally improve the past two years on my Twitter feed. The pace is high, and the game of saying 'Ah, but can they do X' feels like short-term thinking. It may not be long before any image we can describe can be generated in a few seconds.
It's also possible the current generation of AI will hit an upper limit and soon we'll be able to instantly distinguish between AI and human creations.
It feels like a mini-singularity. We sense a big change is imminent, but it's difficult to predict the consequences.
And yet in spite of this, the more AI images I come across, the less I feel moved by them. I find myself wowed by the technical achievement, or hunting for the tell-tale glitches. Improvements in technology are seemingly offset by my own desensitization to AI imagery.
I think this is because I'm no longer looking at them as artworks. There are exceptions, for example Memo Akten's Distributed Consciousness. But it feels significant that here I'm naming not an AI system but a human artist, someone I've met and whose work I know.
In life, I'm bombarded by images in adverts, diagrams, the news. Mostly I ignore these, or at least extract the sentiment and move on. My barriers are up. I'm closed. To be moved by art, I need to be open to being moved.
It's magical to be moved by a work of art. I feel a part of the artist's inner world. I feel their voice. I feel a sense of belonging with others who can also feel this. I connect to a deeper plane of human experience that reveals my own inner life through the inner life of others.
Art that does this to me is a gift from the artist. To open myself to feeling this is a gift to the artist. It's a reciprocal relationship.
It's not always easy to open up to an artwork. There's so much demanding my attention, let alone my heart. It feels vulnerable to open up. Anything could happen. Sometimes it takes the authority of a curator, or the intimacy of a friend, to persuade me to stop and look with openness.
And so, context matters.
Belief in that story matters.
I've experienced something similar in my own AI artwork. In Latent Voyage, I invite people to use their full boy movement to navigate the hallucinations of an AI model.
I trained the image model of Latent Voyage exclusively on photos I've taken myself, about 25,000 of them going back 20 years into my days in the darkroom at high school. The experience is an uncanny journey into my own memory. Fragments of the shapes of university buildings emerge, as does my love of photographing trees, rocks and the sea.
The embodied interaction can be quite captivating in itself. But when I tell participants the model's trained on my own visual experiences, I feel a shift in how they receive it. They're no longer just navigating a hallucinatory machine, but glimpsing the human entangled inside that machine.
For me, art is a human-to-human connection. Craft and beauty are important, of course. But they are like the telephone. If there's no one on the other end, it's just an inanimate object.
I suspect the moment I truly feel the voice of an AI as the voice of an artist in an image is the moment I believe it to be as sentient, conscious, alive with a inner experience to express.
So what of nature? Or fractals?
It's not so different. When natural beauty hits me, I feel like I'm connecting to the universe - the living, conscious, sentient universe. There's a reason they call the Mandelbrot fractal 'God's thumbprint'.
As before, it is upon me to be ready to receive it. Or, as William Blake puts it:
If the doors of heaven were cleansed every thing would appear to man as it is, infinite.
For man has closed himself up, til he sees all things thro' narrow chinks of his cavern.
Of course, even if we don't descend into an endless soma bliss of optimal art, an onslaught of AI-generated entertainment, education and manipulation may be imminent. Many illustrators and stock photographers are facing the loss of freelance income. Or maybe a court will rule it a breach of copyright to train a model on an image without the author's permission. Maybe OpenAI et al. will have to pay royalties to those who created the relevant training data (they're charging 15¢ a pop now so this could be serious income). More on this in a future newsletter...
Olot, 26 Aug 2022
[^1]: Images were generated using DALL·E using the prompt: A colorful New Yorker cover illustration without text which shows people in different positions in an art gallery all staring vacantly into their phone. Award-winning New Yorker cover.
This is Tim Murray-Browne's newsletter. You can subscribe at timmb.com/newsletter.