John Szarkowski, the legendary curator at the MoMA, once described photography as “the act of pointing.” And for the nearly 200 years since its inception, photography has consisted of capturing a visual perspective from the physical world using light — first with light-sensitive plates, then film, then digital sensors. When digital cameras became widely available, many photographers lamented the move away from analog technology but basically Szarkowski’s definition still held: Photography consists of pointing, as a reaction to something that exists in the world.
With advent of A.I. image generators, however, this definition feels obsolete.
Generative A.I. tools can produce photorealistic images, typically in response to written prompts. These images are available for purchase from major stock photography agencies, alongside traditional photos. They routinely go viral before being debunked. They even occasionally win prestigious photography prizes. All if which has reignited a two-centuries-old debate: What exactly qualifies as a photograph?
This is not a matter of etymological nit-picking. Calling A.I. images “photographs” — a practice I encounter often — can add to a sense of disorientation in what already feels like a profoundly disorienting moment. Thanks to the ubiquity of digital cameras, we live in a world that’s already flooded with photographs — more than a trillion are taken each year. These digital images can already be easily manipulated through existing tools, including ones built into your phone. Yet they still have some direct relationship to real scenes and events that have occurred.
Now we face a new deluge of images that, however artful or convincing, are at a remove from the world. A.I. images are typically digital composites of countless existing photographs, so by what definition are they themselves real? No wonder some observers are asking “how can we believe anything we see?”
Aside from very real concerns about the livelihoods of professional photographers, especially those who work in commercial photography, I worry that A.I. image generators may leave society as a whole more vulnerable to widespread manipulation — as presaged by hoax A.I. images of Donald Trump violently resisting arrest or, somewhat more comically, of Pope Francis wearing a Balenciaga-inspired coat.
But for all the negative potential, I can also see a possibility that these developments will start a conversation about — and foster an educated skepticism of — all visual media and the relationship of these images, however they are made, to so-called truth.
Artists, writers and theorists have long remarked on our very human tendency to project slippery ideas about truth onto two dimensional surfaces. In 1921, Franz Kafka was told about a miraculous machine that could automatically take one’s portrait, a “mechanical Know-Thyself.” He offered up his own name for the apparatus: “The Mistake-Thyself.” Kafka was ahead of his time — in Susan Sontag’s 1977 essay “In Plato’s Cave,” she wrote, “Although there is a sense in which the camera does indeed capture reality, not just interpret it, photographs are as much an interpretation of the world as paintings and drawings are.” Each photograph, she argued, is inevitably the product of countless decisions informed, consciously or not, by the photographer’s predilections and biases, as well as the limits and parameters of the technology.
So when I hear some people calling the arrival of A.I. an extinction-level event for photography, I often think of the French painter Paul Delaroche who, legend has it, declared painting “dead” after seeing a daguerreotype, one of the first photographic inventions. Painting did not die; it just evolved into a different kind of artistry, freed from the obligations of verisimilitude.
Photography has arrived at a similar crossroads. So I asked four artists who work with A.I.-generated images — Alejandro Cartagena, Charlie Engman, Trevor Paglen and Laurie Simmons — to talk to me about how they’re thinking about the technology and where we might go from here.
This conversation has been edited and condensed.
Gideon Jacobs: Alejandro, you probably have the most experience of anyone here with documentary photography. How do you feel when A.I. images are called “photos”?
Alejandro Cartagena (photographer and publisher of Fellowship, a site dedicated to elevating photography and exploring “post-photography” imagery): Yes, these images are photographic — in some sense. For example, the computer models understand framing photographically. They understand how to use the horizon. They understand how to frame a portrait based on 180 years of photographic diarrhea. These models are looking at images, and the most predominant type of image out there is the photograph. I believe this kind of technology was inevitable because what else were we supposed to do with the trillions of images that have been generated?
Jacobs: That’s so interesting — the idea that these image generators were somehow a natural next step; that we had to find a way to make the glut of photos useful, otherwise we’ve spent the last century amassing an enormous, useless, garbage pile of visual noise.
Laurie Simmons (artist and photographer): Terrence McKenna once said, “Stop consuming images and start producing them” — which is kind of an interesting take on what I have been doing. My first A.I. prompt was on Sept. 2, 2022, and it was sort of … I saw the earth move! I felt like an A.I. whisperer. But at the same time, it raised so many questions and it caused me to go down two consecutive paths: the path of making my work and the path of trying to understand what was going on with this technology culturally, politically and in a corporate sense.
Jacobs: Many have recognized the use of manipulative tools like Photoshop and digital filters for decades, but I don’t remember those conversations ever being as heated as the current one around A.I. images. It seems really difficult to orient oneself or take a position on A.I. when the landscape is constantly shifting. Laurie, does working with an A.I. image generator like DALL.E ever feel to you like a photographic process? Do the resulting images feel to you like photographs?
Simmons: Not really — but I don’t consider myself a photographer. I’m an artist who uses a camera. I see these A.I. images in this sort of interstitial space between drawings, photographs and sculpture. They exist somewhere I don’t have the language for yet.
Charlie Engman (photographer and director): I am interested in photographic imagery because of its ostensible relationship to reality, truth — or whatever. With A.I., a big criteria for me is how well it is able to make photographic-looking images — I’m not personally interested in systems that make images that look like paintings, illustrations or 3-D renderings; I’m invested in the photographic image because it has some kind of direct through line to a notion of truth. Even though I know that images are not true, have never been true, part of me does believe in pictures. Part of my interaction with photographs is a willing suspension of disbelief.
Trevor Paglen (artist and geographer): The idea that a photograph, in and of itself, can record some kind of truth has always been a fiction. Look at Gustave Le Gray, right from the get-go. Look at spirit photography. It’s not possible to make an unmanipulated image.
Simmons: When I picked up a camera initially, I was interested in the fact that pictures could lie, the camera could tell lies. I was never interested in the truth, which is why working with A.I. is such a natural progression for me.
Paglen: You never trust a photograph, right? I’m less worried that we are going to lose some notion of being able to use images to make sense of the world — because we have never made sense of the world only by looking at images. When we do, we end up in weird Loch Ness monster territory.
Cartagena: Everything is subjective. Everything is a selection of reality, hence not reality — not truth.
Jacobs: The immense size of the data sets and the way the A.I. generators connect language and image — it makes me wonder if these images are the closest humanity will get to some version of idealism, to seeing Plato’s concept of forms. Maybe DALL.E’s output with the prompt of a word like “cute” is the closest thing we’ll ever have to some consensus of what “cute” looks like.
Charlie Engman: I recently had an article about my A.I. work published in The New Yorker, and in it I’d sort of flippantly said, The amazing thing about A.I. is that I can make, like, 300 pictures a day. Of course, people on the internet read this as the death of creativity! What was so interesting to me is that labor — the time invested in the creation of an image — was an assumed metric of value. So if you can make it that fast, it’s not art.
Cartagena: But it was the same when film transitioned to digital. I remember the heated conversations in the photo club where everybody was like, “You can make 300 images on one shoot? That’s not right! That’s not real photography.”
Jacobs: Reactions to large technological leaps often tend to fall into one of three camps: the alarmist camp, which sees the technological leap as unprecedented and negative; an optimist camp, which sees the leap as unprecedented and positive, and then a camp we could call the perspectivist camp, which tries to keep things in historical perspective by assuming the leap is similar to previous leaps in some way — leaps to which society, to some extent, adjusted. So which camp do you each align with when it comes to A.I.?
Engman: I would position myself in that last camp — the realist camp. Obviously, I have embraced A.I. in my work. I’m excited about its uses from a creative perspective. But I do empathize with people having anxieties about it and I think we should look at what those anxieties are.
Jacobs: Trevor, are you feeling optimistic, pessimistic or somewhere in between?
Paglen: Probably none of those! These camps are based on the premise that the development of technologies and civilizational progress have something to do with each other, and I don’t think they do.
Jacobs: Laurie?
Simmons: I’m going to go with Terence McKenna on this one and say, “You don’t know enough to worry.”
Jacobs: Alejandro?
Cartagena: I guess I am a perspectivist — because I already went through a cycle of fear and anxiety during the transition from film to digital in the 1990s. I entered photography right at that moment, when film photographers were going crazy because they did not want digital photography to be called “photography.” They felt that if there was nothing hitting physical celluloid, it could not be called “photography.” I don’t know if it’s PTSD or just the weird feeling of having had similar, heated discussions almost 20 years ago, but having lived through that, and seeing that you can’t do anything about it once the technology is good enough, I’m thinking why even fight it? It’s here.