Is AI Photography Really Photography At All?

Recently, I was scrolling through my Instagram explore page when something caught my attention. It was a photograph of a woman walking down the street with her friends, and it was so striking that I had to click on it. In the photo, the woman is wearing a pink floral dress which clings to her slender body. Her brown hair is blowing in the wind, and her lips are slightly parted as she stares right at us. In the foreground are some out-of-focus flowers, in the background are her blurry, equally slender friends. The photo is in motion, washed over with that feminine, Petra Collins haze so indelible to the early 2010s. As I zoomed in on the woman’s face, I was captivated by her other-worldly beauty. Who was she? I drifted down the comments.

“Who is she?” asked one person. A few others asked the same. “She literally doesn’t exist” said someone in the replies. “i thought these were real :(“ said another. This is when I reached the top comment: “God, I hate AI art so much.”

At the time of writing this essay, the “AI photography” hashtag on Instagram is a digital river of waifish, soft, symmetrical figures washed in glossy hues of blues, yellows, and pinks. These subjects look directly at the viewer, challenging them to accept the veracity of the image. Many of the people who post these images, like the person who created the woman in the floral dress, are themselves photographers. Their pages blend AI generated images and their own photography – but the AI work tends to attract the most attention. The image of the woman in the floral dress received over 92,000 likes.

Photography, as an artform, has always coveted its own proximity to reality: a “truth claim”, as scholars call it. In the first half of the twentieth-century, Andre Bazin insisted that photography had nailed down the human impulse towards realism, arguing that  “Only the impassive lens, stripping its object of all those ways of seeing it … is able to present it in all its virginal purity to my attention and consequently to my love.” Photographers throughout the decades have hinged their success on the authenticity of their images, the souls they entrap in their cameras, their ability to freeze time, the hidden spaces they can pry open and reveal to the rest of the world. The project, for so long, was to embalm the realities – beautiful and ugly, of life on earth.

Yet the project of AI photography, which has no claim to reality, seems to focus on something else entirely. With its smooth lines, symmetrical faces, and glossy veneer – AI photography appears to be operating only in the pursuit of aesthetic pleasure.

An image made by generative AI is an amalgamation of an almost-infinite number of other images scraped from the internet. The faces you’re seeing are composites of thousands of other faces in similar photographs tailored by a given prompt. It’s only natural to assume, then, that these AI photographers may have been prompting the bot to create beautiful images for them. So I did my own investigating, and downloaded one of the most popular AI image generators, Midjourney. While it’s recommended to use the most specific language as possible to harness the best results, that would not give me the answers I needed. Inputting the most vague language I could into the prompt, I searched for its default. “A photographic image of a woman in a pond surrounded by flowers” gave me four images of pale women, with elegant faces and thin waists, traipsing amidst a bed of orange, red, and white flowers. Maybe flowers were too indicative of beauty. I went for something even vaguer. “A photograph of a woman walking down a busy street with friends behind her. She is looking at the camera.” Again, four images of a thin white woman with a slender waist and delicate, symmetrical face. One of the images, without my prompting, featured a woman in a low cut shirt and heavy cleavage. On my last attempt, I went for something a bit different. Maybe if I prompted the bot to mimic the work of a photographer known for capturing imperfection, it would change its tune. “Photograph in the style of Diane Arbus, of a man sitting on a curb in New York.” The four men it gave me, again, had chiseled, symmetrical faces.

While the image of the woman in the floral dress is strikingly, almost hypnotically, beautiful, it’s difficult to reconcile its claim as a photograph. These images mimic the language of photography – depth, focus, texture, light, motion, and a hyperrealism almost completely indiscernible to the untrained eye. But they did not happen. What’s more, their appeal seems to stem from an achievement, in my mind, in aesthetic perfection. I clicked on the floral dress image because it was arrestingly beautiful.

Maybe AI photography has more in common with a painting than with photography. When Lev Manovich first came across the concept of digital cinema, he found that the moving image was “no longer a kino-eye, but a kino-brush,” in the sense that the digital image was comprised of a bunch of pixels, which could be altered and substituted for one another (Manovich 20). Instead of reflecting reality itself, the digital image created a new kind of realism, which is “something which looks is intended to look exactly as if it could have happened, although it really could not.” Programs like Midjourney allow users to refine the image in order to land on something more tailored to their desired effect. But it seems that the bot itself undergoes a bit of a refinement process, taking a large dataset based on a general prompt like “woman” and, like a painter with their brush, shaping the face down to a perfect, almost Pythagorean configuration of Western beauty.

But maybe these AI images are not the disruption of photography that they threaten to be. Way back in 1977, before digitization was a fully realized thing, Susan Sontag was already questioning the “truth claim”. She argued that, despite that authentic and transparent quality that gives photography its status as an “objective” art form, the photographer is still operating in the “shady commerce between art and truth.” She says, “Even when photographers are most concerned with mirroring reality, they are still haunted by tacit imperatives of taste and conscience.” Ultimately, for Sontag, photography interprets the world no differently than a painting would.

The AI photographers often caption their photos with promises of a past or future truth: “concept based on something I’m working on” or “AI made with my own images”. They seem to use AI as a visualization tool. Some sort of digital assistant for enhancing the photographic image – pushing it to its maximum aesthetic potential. And the results they get provide a fascinating, if not pernicious answer: maybe truth is less alluring than pleasure.

As featured in SLEEK 78 – BLISS. Available in print and digital here.

Maia Wyman

Maia Wyman, better known as Broey Deschanel, is a video essayist who explores media analysis, cultural studies, critical race theory and politics. Alongside producing videos for YouTube, CBC, Mubi and Nebula and her own channel, Wyman co-hosts the podcast Rehash which discusses the fleeting social media phenomena that hit a nerve in our culture.