There was a year when I took ten thousand photographs a month.[2] Back then, the cost of each picture was time spent walking: hours at the lake park, around Seoul, and at historic sites such as Gyeongbokgung. Ten thousand pictures meant almost all my time outdoors that month. One picture might seem cheap. Ten thousand were not.
With generative AI apps, that cost disappeared. I no longer walk to get an image. I sit and type words. Hundreds of images, or more, can come out in an hour. Before, I paid for each picture in distance. Now I pay almost nothing. The cost of taking the picture has gone.
What happens when the cost disappears? It does not disappear. It moves from making to choosing and arranging. When an image costs nothing to make, a hundred appear. Deciding which to keep and how to arrange them brings the whole cost back. The time I once spent walking is now spent choosing and arranging. The weight of the work has shifted.
I wrote earlier about reducing tens of thousands of photographs to thirty.[1] That was about how difficult choosing can be. Here I am looking at something else: why has choosing become so costly? Because making has become cheap. They are the two pans of a balance. When one goes down, the other goes up.
Photography has been through this before. With film, each frame cost money. A roll held thirty-six exposures; when it was finished, you bought another. So you thought before pressing the shutter. With digital photography, each frame became free, and people pressed instead of thinking. Choosing became, for the first time, a major part of a photographer’s work. I learned photography after that change, so I began as someone who had to choose. Generative AI repeats the change, on a larger scale.
I see this in class. In photography, we had to change the framing and the lens, working hard to leave things out. With generative AI, one word, “minimalism,” can do that work. Making has become that much cheaper. But then students get stuck. At first, they are amazed by the beautiful images they can make as soon as they learn. Once they become familiar with the process, they produce dozens and no longer know which to keep. Everything looks convincing. If everything looks good, they cannot choose. Where the cost of making reaches zero, the cost of choosing and arranging becomes visible for the first time.
I leave almost all the photographs I take on a storage drive without posting them. Only the ones I choose get renamed and stored separately. Renaming is the first step in choosing and arranging. When I remove the camera’s number and add a date, a place and a title, the picture moves from something made to something chosen. Generated images need the same attention. Far more are waiting for it than when I took ten thousand photographs a month.
At the Martin Parr retrospective, some five hundred photographs selected from fourteen series were on display.[3] Standing before them, I wondered what fraction of everything he had photographed over fifty years had reached those walls. There must have been more photographs left out than shown. That ratio is the cost of choosing. For a photographer’s work to endure, the cost may lie not in taking the photographs but in leaving the unshown ones unshown.
People say making has become easier. They are right. But the work has not become easier. Its cost has not fallen; it has moved. For someone who cannot see where it has moved, the work looks free. For someone who can, it has become more costly than before.
When the cost of getting an image reaches zero, what has the cost of keeping one become?
Originally published in Korean on October 11, 2026, as 「만드는 값과 고르는 값」.
1. The earlier account of reducing tens of thousands of photographs to thirty: “From Tens of Thousands to Thirty” — 「수 만 장에서 서른 장으로」 (in Korean).
2. The year of taking ten thousand photographs a month: “Ten Thousand Photographs a Month” — 「한 달에 만 장」 (in Korean).
3. Exhibition reference: Seoul Museum of Photography, Martin Parr: We Are Martin Parr, July 16–October 18, 2026.
