In the Age of AI, Photography Is Now Untethered

In his short story “The Adventure of a Photographer,” published some 70 years ago, Italo Calvino presciently described our obsessively photographed world from deep within the era of analog film. He writes, almost half a century before the advent of today’s omnipresent camera phones, of droves of city dwellers likened to hunters who, as the weather becomes warmer, go out on Sundays photographing one another before returning to anxiously await the arrival of the developed pictures.

‘The Synthetic Eye’ by Fred Ritchin

$30

Thames & Hudson

These people’s lives, the story suggests, are unfulfilled until the photographs are before them, their experiences remaining otherwise somewhat vague and abstract. Not until they receive the pictures do they appear “to take tangible possession of the day they spent”; it is only then that their views of a landscape, a child at play, even the light reflecting upon a spouse, are concretized and validated, no longer subject to the vagaries of memory.

Calvino’s story, written in the 1950s, was translated into English and published in full in Vanity Fair in 1983, the year the magazine relaunched as a monthly. Its protagonist is Antonino, a non-photographer. He decides nonetheless to join the others and, after meeting a woman, Bice, tries to capture her essence with his camera. Aware of the many ways in which he could photograph her, as well as those versions of her that he cannot extract with his camera, he seeks a strategy to combine the two. As a result, he decides to photograph her constantly, day and night, intent on exhausting all possibilities.

It also exhausts her. Antonino photographs Bice nude and tells her, “I’ve got you now.” She bursts into tears, bewildered, and soon leaves him. He falls into a serious depression, yet, newly fixated on what the camera can convey, begins to keep a photographic diary and, while not leaving his home, devotes much of his time to staring vacantly ahead while photographing her absence. Then, Antonino looks at what he has done, lays out the imagery on top of newspapers, and photographs the entire ensemble, juxtaposing public and private pictures in a way that resembles the chaotic mixture that populates the Internet today. He concludes that perhaps a “true, total photography” consists of this private imagery being displayed “against the creased background of massacres and coronations.”

Finally, progressively more frustrated, Antonino comes to an epiphany. He understands “that photographing photographs was the only course that he had left – or, rather, the true course he had obscurely been seeking all this time.”

Antonio’s long-ago method is not that different from the way image generators work today: artificial intelligence systems train on an enormous archive of online photographs and contextualizing information to generate their own photo-like creations based on existing photographs. If Calvino’s tale were written now, Antonino could continue ad infinitum, creating new images—synthetic ones—resembling Bice nude, clothed, young, or old, making love with him or with someone else, or using previous photographs of her to generate deepfakes that place her, without her consent, in situations where she never had been.

We humans make ourselves minor deities of a world that does not exist, while remaining powerless to affect substantive change in the actual world we inhabit.

Or Bice, should she want to take more control of her own image, might be making and distributing endless selfies with her cellphone while using software filters to enhance her look, a self-absorption that can be, just like Antonino’s stalking, obsessive and ultimately unhealthy. As Monica Lewinsky, a longtime advocate against online bullying, once put it, “The whole point of a filter is saying: I want to look better, which means you’re not good enough.”

Unlike in the 1950s, it can be much more difficult for images, no matter how exploitative, to be forgotten and erased. Today, the image can manifest in many more ways, some of them highly destructive. The actress Scarlett Johansson, whose likeness has been utilized in pornographic deepfakes, forcefully characterized the situation: “The Internet is a vast wormhole of darkness that eats itself.” Jennifer Lawrence, the victim of a nude-photo hack when she was 24, told Vanity Fair: “It is a sexual violation. It’s disgusting. The law needs to be changed, and we need to change…. These Web sites are responsible.” (Her perpetrator pleaded guilty to hacking into her and other celebrities’ personal accounts and removing private information that then found its way onto the Internet. He was sentenced to 18 months in prison for a felony violation of the Computer Fraud and Abuse Act.)

In the 1990s, as Photoshop became available to the public at a relatively affordable price, photographic alterations began to be reconceived as necessary enhancements rather than deceptive manipulations. Since then, the modification of photographs has become commonplace, the photograph itself considered more of a first draft to be altered than an authoritative rendering of what was in front of the lens. If “the medium is the message,” as media theorist Marshall McLuhan advised, the message of photography today in the digital environment is the malleability of the real, not its recording as something fixed for future contemplation.

“completely change your pictures with AI.” And by doing so, one can radically change the look of the world to match the image creator’s preconceptions. Rather than observing, this kind of computational photography advances a form of proprietary looking, an act by which what we want to see is realized as a form of consumer entitlement, an eighth day of creation with outcomes and enhancements guided by camera manufacturers and software developers.

The image, no longer maintaining a dialectical relationship with the world, is often constructed according to fantasies of beauty and happiness, a map of desires allowed to supersede the realities that photographs have traditionally explored. Nor does the photographer necessarily control the results. As Stephen Shankland described it for CNET: “With a sweep of my mouse, Photoshop could generate a nice patch of blue sky to replace an annoying dead tree branch cluttering my shot of luscious yellow autumn leaves. Smartphones are now making similar decisions on their own as you tap the shutter button.”

As a result, we humans make ourselves minor deities of a world that does not exist, while remaining powerless to affect substantive change in the actual world we inhabit. This conundrum has perilous implications in the current age—politically, economically, and spiritually—given how little we now know about what is going on elsewhere, as our established norms, institutions, and assumptions are shattered. On an even more basic level, AI-generated images simulating photographs invoke so many convincing scenarios that the actual ones become no more real than any other.

Images have now become the world’s lingua franca. There are so many – one estimate put the number at around 5 billion photos produced daily, with an average of 2,100 to be found on each smartphone, along with 3.37 billion people watching videos online – that individual images inevitably lose their significance.

Should all of these billions of still images still be called photographs? Stephen Mayes, previously the recording secretary of World Press Photo, a global organization that awards annual prizes to photojournalists, described this transition a decade ago in Time in an article called “The Next Revolution in Photography is Coming.” Mayes wrote, “Digital capture quietly but definitively severed the optical connection with reality, that physical relationship between the object photographed and the image that differentiated lens-made imagery and defined our understanding of photography for 160 years. The digital sensor replaced [the] optical record of light with a computational process that substitutes a calculated reconstruction using only one third of the available photons. That’s right, two thirds of the digital image is interpolated by the processor in the conversion from RAW to JPG or TIF. It’s reality but not as we know it.”

Regarding the Pain of Others in Israel and Gaza: How Do We Trust What We See?Arrow

Complicating it further, Shankland of CNET explains, “smartphones today composite several frames – up to 15 in the case of Google’s Pixel 8 Pro’s HDR technology – into one photo. Stacking multiple frames lets the camera handle shadow detail better, reduce noise, and show blue skies as blue, not washed-out white. But it also means that one photo is already a composite of multiple moments.” In effect, many of the latest digital cameras render obsolete the highly celebrated dictum by Henri Cartier-Bresson that in photography there is a singular “decisive moment” in which the photographer recognizes and captures “the significance of an event as well as of a precise organization of forms which gave that event its proper expression.”

Mayes concluded that manufacturers, needing to camouflage these differences to market their digital imaging devices as a form of photography, made their images imitate photographs, making it possible, “as long as there’s an approximate consensus on what reality should look like, [to] retain a fingernail grip on the belief in the image as an objective record.” Or as Geoffrey A. Fowler posited in the Washington Post, in an article entitled, “Your smartphone photos are totally fake – and you love it”: “Think of your camera less as a reflection of reality and more [as] an AI trying to make you happy.”

These illusions are pushed even further by artificial intelligence systems that generate photorealistic images from scratch without the use of a camera. “Now we’re in a reality where people just choose their history,” declares Phillip Toledano, an artist who recently worked with artificial intelligence to reimagine the United States of the 1940s and ’50s. “I think it’s an extraordinarily important point in history where we’ve come to the death of truth,” he told LensCulture. “Every lie can now have convincing evidence. And so that’s what this work is showing: look how convincingly we can create history that never happened with this technology. I think that as a species, as a society, we’re going to have to figure out a new way to understand what is true or what’s not. Or it may be that we’re entering a moment in history where we accept that there is no visual truth anymore.”

For all these potential pitfalls, it should be emphasized that artificial intelligence systems, if freed from the limitations of visual realism, can also be extremely useful tools, drawing on their vast banks of accumulated data to circumvent truisms or challenge stereotypes, opening up a wide, hybridized range of possibilities based on diverse sources and conceptual approaches, including some that have fallen out of favor. The resulting imagery, for example, may elicit a quantum worldview over a Newtonian one, exploring parallel universes rather than a more conventional representation of cause and effect. Instead of destabilizing photographic credibility, AI-generated imagery has the ability to explore more far-flung realms, expanding our minds and, perhaps, positively shaping our destinies.

The role of photographs has moved on significantly from what Calvino put forth 70 years ago, when he describes them as eagerly awaited to concretize and validate one’s experiences of the day. Now, increasingly, the camera’s recordings are transformed by software, enhanced and upgraded into more palatable imagery, describing a day that never occurred.

Fred Ritchin is dean emeritus of the International Center of Photography, former picture editor of The New York Times Magazine, and author of The Synthetic Eye: Photography Transformed in the Age of AI.