This is a book about doing something good, once, and trying to do it again, and forgetting how hard it was to do it the first time.
Sure, it’s about a lot of other things, too. It’s about memory, and laziness, and artificial intelligence. It’s about the climate crisis, and the pandemic, and growing up in a world forever changed. It’s about seeing the forest for the trees, or, rather the skies for the birds. And, I suppose, it’s a book about pigeons.
I’ve taken a lot of pictures with birds in them. Most of these aren’t pictures of birds, although some are. No. It seems, looking back at my past work, that birds seem to wander into my photographs, soar through the gaps in my compositions, at just the right moments.
Looking back at these old pictures, they feel like miracles. That’s what I remember about them. Stumbling on incredible light: jubilation. A pigeon zips across the space between two trees, like an arrow shot by gods. Miracles occur, and of course they do. A miracle is too perfect to not happen.
It’s easy to forget there’s never a guarantee.
Successful photographs are made to look effortless, uncontrived. A good work of art can be understood, writes the photographer Robert Adams in his 1981 essay Beauty in Photography, “by the apparent ease of its execution… Only pictures that look as if they had been easily made can convincingly suggest that beauty is commonplace.” Looking at one’s old work, it’s easy to buy into the world they suggest, where miracles (birds, light, punctum) are abundant, where cameras are magnets for beauty. It’s easy to forget all of the waiting, the watching, the time spent staring up at the clear, empty sky. It’s easy to look at effortless pictures and expect that the next picture will take just as little effort.
But, of course, good art is difficult. Enter artificial intelligence.
Where real art requires time, hard work, and risk, AI art offers a glittering shortcut. Skip the blank canvas, skip the empty sky, and go straight to the product. All you need is a Big Idea.
Big ideas come easy - fill the sky with pigeons! - but big ideas are not art. Art is a tangible artifact, wrought by hours, weeks, years of tiny, tough decisions. Photographs start as unexposed film. Poems start as empty pages. Paintings start as blank canvases. Every subsequent shot, word, or brushstroke narrows down what that emptiness can become. Each microscopic decision is deliberate, a part of the whole, a bold statement against entropy. Art, in short, is the product of a practice, not a prompt.
Even in this short essay, I’ve made hundreds of decisions. The decision to include this paragraph was difficult. Is it necessary? Is it implied? Are rhetorical questions cliché and overused? Cliché and overused are basically synonyms, but they sound good together. Whatever I choose, the reader won’t know the alternative. Which one better suits the piece as a whole?
For the AI artist, all decisions are multiple choice. The AI artist is never confronted with the blank canvas. They are never forced to wait, to deliberate. Their decisions are all broad strokes: Should I ask for a blue pigeon, or a green one? Which version of the model’s output do I like better? Which real artist’s style do I want it to mimic?
Generative AI encourages complacency. The microscopic decisions that a real artist makes are hidden from the AI artist, in billions of unintelligible parameters. What does the AI artist contribute? If art is the communication of the human experience, what experience does AI art communicate? Laziness? Instant gratification?
The impetus for this book was in May, 2024, when I realized there were a number of invisible standards against which I judged my pictures. For example: the presence of birds. If a picture includes any stretch of the sky, it is certainly better off with wings in it, for dynamism, for the decisive moment, for a touch of punctum. As such, I started to disregard any picture that did not have a bird in it. My standards went up. Miracles became requirements.
Thus, so many of my photographs were immediate failures. I began to wonder: are my pictures actually better than they were before? What would an obsession like this look like, taken to its logical extreme?
I used Photoshop’s generative fill tool to add pigeons to empty skies, and I started naming them. I assembled four groups of names: nice names, culture names, science names, and gibberish names.
The nice names were easy. What’s more natural than a pigeon named Pauline? When I needed inspiration, I looked to my neighbors, and to local street names. I found names in books like Bolaño’s The Savage Detectives and Márquez’s One Hundred Years of Solitude. During the Paris Olympics, I named pigeons after competitors in balance beam, trampoline, triple jump, and other bird-like sports.
The culture names came next. More explicitly referential, they include the names of painters, poets, historical figures, characters from Shakespeare and other literary greats. Names of the zeitgeist. Laziness sets in. Why work for original names when you can cash in on cultural cachet?
Then, other names: computer scientists, tech CEOs, AI assistants. They descend into gibberish. In a transcript of a conversation with an AI chatbot, I found the line: I am Looking For a God and I’ll pay you for it. From a worship of creativity to a worship of creatives to a worship of the automation of creativity, we settle on complacency. Easy worship. Hollow art. Why work for beauty when you can pay for it?
Originally, this was going to be the end. A descent into madness. No hope for the AI artist. But to my surprise, when I used AI image tools for the first time, the process was anything but enticing. It was fickle, unwieldy, and incredibly frustrating. I found myself exaggerating all of my prompts. If I wanted three pigeons, I had to say I wanted a flock. If I wanted a flock, I requested a hundred billion pigeons, an unfathomable overabundance of pigeons, and even then, it often only gave me three. If I wanted big, I said huge. If I wanted small, I said infinitesimal. It was like performing surgery with a backhoe.
Even in a project about AI art’s inadequacy, I had trouble making images that belonged. And this gave me hope. The sky, in fact, has not run out of pigeons. They are still there, in trees and under benches and huddled up on chimney pots. Occasionally, they take flight.
Sometime in the last few years, I forgot that miracles take time. They wouldn’t be miracles, otherwise. Good pictures suggest a world in which beauty is commonplace, yes, but they suggest something else: that I found beauty, once, and I can do it again.
