Editor’s note: Images throughout this article were created by photographers in Chris Buck’s Checking Focus group in response to the assignment “Make an image that AI can’t.”
Photography is having a moment.
In a recent advertising campaign titled “What Shapes Your World?” The Globe and Mail, Canada’s national newspaper, leans on the visual hallmarks of traditional photography to evoke what makes reporting trustworthy. The spot lingers on strips of film negatives examined in a red-lit darkroom and contact sheets marked with grease pencil selections, and accompanied by a gravelly voiceover: “fact, context, clarity.” Anyone who has spent time in a darkroom recognizes these as the material artifacts of photojournalism at work, and seeing them revered carries a certain emotional charge. They evoke a time when photographs were understood as more than the fleeting images that flash across our screens today. They were the hardcopy witnesses: light physically striking film, moments unfolding frame-by-frame, editors selecting what would ultimately represent reality to the public. The newspaper’s message centres on photography’s ability to bring readers what we are starving for now: credibility and context.
The symbolism is deliberate. At a moment when generative AI can produce convincing images from a line of text, news organizations are increasingly emphasizing photography’s association with clarity, verification, and presence. In her recent From the Editor newsletter, Toronto Star Editor-in-Chief Nicole MacIntyre reflected on why every cub reporter should “make friends with the photogs.” She describes photojournalists as among the people who know a community best, often arriving first, sensing when a situation might turn, and helping reporters navigate unpredictable scenes while documenting them for the public. The profession, she wrote, tends to attract people who are “fearless and fierce,” part artist and part first responder.
The emphasis is not limited to visual journalism. In a recent audio advertisement circulated on The New York Times’ podcasts including The Daily, publisher A.G. Sulzberger urged listeners to support “any news organization dedicated to original reporting.” He emphasized the importance of facts and context gathered by journalists who are physically present at the scene of events. In a media environment where deepfakes and fabricated broadcasts can move widely online, the value of journalism increasingly rests on something both simple and difficult to replicate: being there.
Against this backdrop, portrait photographer Chris Buck recently posed a deceptively simple assignment to the photographers in his mentorship group. He asked participants to “make an image that AI can’t.”
The prompt was not meant as an anti-technology provocation, Buck told me, but as a way to ask a deeper question: What, exactly, still makes a photograph human?
Buck has spent decades photographing cultural figures for publications including The New York Times Magazine and Vanity Fair, developing a reputation for portraits that are uniquely his own. They are carefully composed and slightly unpredictable, where something unplanned slips into the frame.
He writes a newsletter called Uneasy, and in recent years he has begun teaching a workshop called The Surprising Portrait. He also hosts an ongoing critique and mentorship group called Checking Focus. It was within that group that he issued the assignment.
“Make an image that AI can’t,” Buck wrote to participants, asking photographers to produce “a surprising and very human portrait” that artificial intelligence could never generate.
When the group met online to review the results, the discussion quickly moved beyond technical tricks and into what makes a picture human.
“What AI can do is weird,” photographer Jonah Atkins observed during the critique session. “But it’s not motivated weirdness, it’s weird in a way that makes absolutely no grounded sense.”
Artificial intelligence can generate strange imagery, distorted faces, surreal juxtapositions, dreamlike compositions, but the strangeness doesn’t emerge from lived experience. It is not grounded in a relationship between photographer and subject. There is no accidental gesture that defines a compelling portrait, like Churchill’s scowl when Yousuf Karsh famously removed his cigar.
Human photographs, Buck suggested, tend to contain small, unresolved moments: a partial expression, an ambiguous gesture, a visual element that doesn’t fully explain itself.
AI, by contrast, tends to resolve things too neatly.
“It wouldn’t be okay with that kind of ambiguity and sense of spontaneity,” he said.
Several photographers, including Andrew Fleming, responded by turning the camera on themselves, creating self-portraits that conveyed vulnerable and complicated inner states.
“It can’t do you,” Buck said simply.
Among them was Toronto photographer Alex Franklin, who had first met Buck during one of his The Surprising Portrait workshops in Toronto. After seeing her work there, Buck later invited her to intern with him in New York, and she now participates in Checking Focus alongside other former workshop participants.
Franklin’s photographs for the assignment emerged almost accidentally. While staging a portrait of her boyfriend on a worn couch, she accidentally knocked over a nearby plant, spilling soil across the cushions.
Instead of cleaning it up, she incorporated the mess into the image.
“I made em’ lay it in,” she said during the group critique.
In our conversation and in her own writing about the series, Franklin described the moment as unexpectedly revealing. The couch had already become a metaphor for the period she and her partner were in: two young adults, as she put it, “trying to pull ourselves out of a rut,” quite literally sinking into the furniture. When the soil spilled, her first instinct was frustration. Then she realized the accident belonged in the photographs.
“This is real life. Not perfect… but so human,” she reflected.
The resulting portraits leaned into the imperfect intimacy of the moment, something she felt artificial intelligence could not meaningfully replicate or understand.
For Buck, that quality was immediately visible.
“The level of intimacy and subtlety, it’s just something AI cannot do,” he said.
Other photographers approached the assignment differently. Atkins recreated a family snapshot inspired by an old photograph of his great-grandfather, drawn to the fleeting expression in the eyes. Bashir Abdulkareem sought subjects whose distinctive features, like vitiligo, resist easy replication by generative systems due to the lack of existing training data.
Many of the strongest images shared a common characteristic: ambiguity. Matt Coughlin presented faces partially obscured, gestures interrupted mid-action, details entering the frame in ways the photographer could not fully plan.
For Buck, those moments are often where photographs become interesting.
“It could be just a little glint in the eye,” he said. “Or the way they turn their head… that makes something pop.”
Those small accidents, the unpredictable results of human interaction, are impossible for generative systems to replicate because they’re not in the room with their subjects.
When synthetic images blur our concept of reality, the distinction between what is generated and what is experienced becomes more consequential.
Buck was careful not to frame the exercise as a rejection of artificial intelligence. Like many photographers, he uses AI tools himself in limited ways, sometimes to sketch ideas or mock up concepts for clients.
“This wasn’t meant to be an anti-AI assignment,” he told the group. “It was meant to be prompting thought.”
In fact, the discussion often revealed as much about photographers as it did about technology. Participants were discovering what aspects of their work they considered essentially human.
In some cases, like Franklin, that meant leaning into imperfection. In others, it meant embracing ambiguity, in photographs that leave something unresolved.
Or, as Buck put it in our conversation, moments where the image contains just enough uncertainty to feel alive.
In journalism, that element of presence has always mattered, because it turns image into evidence.
The next assignment Buck gave the group built directly on that idea.
Quoting jazz musician Thelonious Monk, he framed it this way: “A genius is the one most like himself.”
The photographers were asked to generate five ideas that felt impossible or rarely attempted, then create an image exploring one of them. Buck jokingly described the exercise as “AI assignment part two.” But the premise was serious. If artificial intelligence can produce images in almost any visual style, the value of photography may increasingly lie elsewhere: in the experiences photographers pursue, the relationships they build with subjects, and the unpredictable moments that occur when someone stands in front of a lens.
When synthetic images blur our concept of reality, the distinction between what is generated and what is experienced becomes more consequential.
In journalism, that element of presence has always mattered, because it turns image into evidence. It is the record of an encounter, a photographer present in a place witnessing events unfold and deciding when to press the shutter.
Artificial intelligence may become extraordinarily good at producing pictures. What it cannot do is stand in a room with another person, share a moment of tension or hesitation, and capture the small, unplanned gesture that reveals something real.
In the end, the most meaningful distinction may not be the image itself, but the fact that someone was there to make it.
Chris Buck’s next The Surprising Portrait workshop will take place in Montreal, June 12–14.
The weird part I’m trying to make peace with is being “okay” with the reality of an audience who may not value or care for the fact that someone was there to make the image. Is that a battle worth fighting? Is it even a battle?