two years ago, a photographer in logan square told me she stopped listing her work on stock sites. the income had already dropped 40 percent. by the time the ai flood really hit... it wasn't worth logging in. "why would anyone pay for my portrait session," she said, "when they can generate 200 variations of a face in 30 seconds?" i understood the despair. i also knew she was asking the wrong question.
the truth is... nobody tells you this but the market for generated images did not steal from human creators. it exposed them. it revealed something that was always true: undifferentiated digital work has no inherent value. there is no scarcity in a thousand similar stock photos. there never was.
what the ai flood actually did was create the opposite problem. it generated so much visual noise that human-made work became legible again. not because it was beautiful. because it meant something. because somebody made a choice.
when quantity destroys itself
the numbers tell a specific story. in 2022, human-generated image listings on major platforms held steady around 600 million assets. by mid-2024, ai-generated images crossed 2 billion. stock prices for standard digital work dropped 60-75 percent across most categories. licensing revenue for photographers and illustrators fell harder than anyone predicted.
but here is what the panic missed: the creators who actually made money were not the ones competing on volume. they were the ones who could prove intent.
a wedding photographer in pilsen told me her rates actually went up. not because her photos got better. because a bride now knew that when she hired someone, she was paying for decisions. for the moment someone chose to shoot at golden hour instead of letting algorithm optimize for brightness. for the risk of showing up to a venue and trusting instinct about light instead of computing 10,000 variations.
a portrait artist i know in wicker park started signing prints with process notes. not artist statements. actual technical choices. "shot on kodak portra 400, prime lens, single light source, one take." the work was the same. the framing changed everything. suddenly it was not just an image. it was proof that a human had limitations. had committed to something real.
this is the thing ai cannot fake: the cost of failure. when you generate 200 variations, you are not choosing anything. you are exploring. when you shoot one roll of film and develop it, you are exposed. the choice becomes visible. that visibility is what people actually want to pay for.
the signal problem nobody expected
before the flood, digital work had a different problem. there was too much. endless scrolling. no way to know if something was made by someone with taste or just someone with technical skill. the market could not distinguish between intention and competence.
ai solved that by accident. it made the distinction vivid. now when you see work, you are asking: did someone choose this or did a system suggest it? that question matters in ways we are still figuring out.
i watched a designer in logan square pivot completely. she stopped selling templates. started offering process consultation. "show me your brief," she would say. "then i will show you what i would make and why." she charged three times what templates cost. got booked four months out. the difference was not her skill. it was that she was now selling the reasoning, not the asset.
a musician friend uses ai to generate harmonic variations when she is stuck. but the work she releases, the work she charges for, is the version where she made the hard choice about which variation to develop, which one to strip down, which one to leave broken on purpose. the ai did not write the song. it helped her hear options. but the final answer... that is all her.
what changed is the market now rewards you for being willing to be wrong. to commit to one direction instead of hedging with variations. to say "this is the choice i made" instead of "here are 200 options, pick one."
that is the opposite of what most creators thought would happen. they thought ai would make choice irrelevant. instead it made choice everything. because in a world of infinite generation, the only thing scarce is somebody actually deciding something matters.
what this means for the grind
if you are out there building work right now... the old rules are dead but not how you think. you cannot compete on speed or volume anymore. but you also cannot hide behind craft alone. the thing that pays is the ability to articulate why you made it this way and not that way.
this is harder than it sounds. it means knowing your taste. developing it. defending it. failing at it and trying again. it means showing the work, not hiding it. it means saying "i tried five approaches and chose this one because..." and having a real answer that is not about technical perfection.
the creators i know who are thriving right now... they are not better than they were two years ago. they are clearer. they have stopped trying to make work that pleases everyone and started making work that proves they believe in something. that distinction is now worth money. for the first time in a long time, it actually is.
the ai flood did not kill digital art. it killed the idea that visibility equals value. it killed the stock photo dream. it killed undifferentiated work at scale. what it created instead is smaller and harder and infinitely more honest... a market where what you make only matters if you can prove you meant it. and in a world drowning in generation, that proof is the rarest thing there is.