So AI Made It? Maybe We're Asking the Wrong Question.

I keep noticing the same reaction whenever AI comes up in art:

"AI."

And somehow that already sounds like a verdict.

AI cover? AI. AI-assisted ad? AI. Digital artwork made with AI? AI.

And the conversation usually stops right there.

The idea seems to be: if AI touched it, the artist didn't really do anything.

This surprises me. Not because I think every use of AI is automatically artistic or fine. But because it seems like people've forgotten something pretty basic about art: the tool and the artistic act are not the same thing.

Take Yves Klein. In 1960, for his Anthropometries, he didn't paint the canvases himself. He covered female models in his patented International Klein Blue and directed them to press their bodies against the paper. He called them "living brushes." Klein stayed physically removed from the whole thing. The models did the actual mark-making. And yet nobody says the works were "made by the models." The idea, the color, the setup, the rules, the performance, the decisions — all Klein's. The physical execution was delegated.

Or take Sol LeWitt. Most of his wall drawings were never touched by his own hand. He'd write a set of instructions — sometimes just a line or two, like "ten thousand straight lines at random" — and hand them to a team of draftsmen who actually drew them on the wall. LeWitt himself said it plainly: the idea is the work, and whether or not the artist physically executes it "makes no difference." Nobody looks at a LeWitt wall drawing and says the draftsmen made it.

This isn't new either. Photography went through the same fight. In 1857, the critic Lady Eastlake argued that photography couldn't really be art, because it was too mechanical — the camera just copied reality, no real creative judgment involved.

Now we don't think twice about a camera. Or Photoshop. Or scanners, 3D software, printing tech, any of it. We call them tools.

So why does AI get such a different reaction?

Part of it is just psychology. A study in Scientific Reports (six experiments, almost 3,000 people) found that people rated the exact same artwork lower the moment they were told AI made it — even when they admitted they couldn't tell it apart from human work, and even when they knew a human was involved too. Another study with over 1,700 participants found something similar: once people believed a piece was AI-made, they saw it as less creative and felt less genuine awe looking at it, regardless of how it actually looked. And when researchers looked specifically at art experts, the bias was even stronger than with everyday viewers — the more someone knew about art, the harder they judged the AI label.

There's also a study of 432 people that I found telling: they thought effort and creativity mattered for AI art, but didn't think traditional artistic skill was really necessary for it. And when it came to authorship, they leaned toward crediting either the AI itself or the artists whose work trained it — rarely the person who actually directed the process.

Which brings me to the real issue. There's a genuine ethical question here. These models were trained on huge amounts of human-made work, often without the artists' consent. "It's just a tool" doesn't make that go away, and I don't think it should. A tool can be useful and still raise hard questions about how it came to exist.

But that question is separate from whether someone who uses AI has "stopped being an artist." The question I actually find interesting is: what did the person contribute?

This is something I've experimented with myself — not as a philosophy behind everything I make, but as a way to actually understand the process from the inside, rather than theorize about it from a distance. What I was after was authorship — making sure that if AI touched something, it still ended up feeling like mine. So I ran some of my own tests: prompting, iterating, training Lora, seeing what the tool could and couldn't do with a visual idea.

What surprised me wasn't how easy it was. It was how hard it was to get something that felt right. I'd spend hours changing prompts, trying variations, throwing most of it away, chasing one specific visual idea. And even then, I still go back over the image by hand. The AI never gives exactly what's in your head. There's a lot of thinking and deciding between an idea and something worth calling finished.

That's why the "AI = no real work" idea feels so off to me. Say you use AI to see how an animal moves because you don't know its anatomy well — have you stopped being the artist? Or have you just changed where your reference comes from? Thirty years ago that might have been a photo. Further back, it would have meant going outside to actually watch the animal. The tool changed. The problem being solved didn't.

None of this means skill doesn't matter — quite the opposite. A photographer needs to understand a camera. A painter needs to understand paint. Someone using AI needs to understand what the model can and can't do, how to push it, how to judge what comes out, and how to turn that into something intentional.

There's a real difference between:

"I typed a prompt and picked the nicest result."

and

"I had an idea, used AI to explore it, made hundreds of small decisions, rejected most of what came out, reworked what I kept, and finished it myself."

The final images can look almost the same. The process behind them isn't.

I think this is what we're still struggling with. We're used to judging artistic work by what we can see — the brushstrokes, the hours spent drawing, the visible effort. AI hides a lot of that labor. And because we can't see it, we assume it isn't there.

But the real work of an artist was never only the movement of a hand. It's deciding what should exist at all. A point of view. Taste. Knowing when something is right and when it isn't.

AI can generate an endless number of images. It can't give any of them a reason to exist.

So maybe the question isn't "was AI involved." It's "what did the human actually do with it."

If the answer is nothing, I have very little interest in calling that an artistic practice.

But if AI became one tool inside a much bigger process — thinking, choosing, directing, editing, making something genuinely personal — I don't see why that stops being art just because the tool is new.

We've rejected new tools before. Maybe the problem was never that AI makes art too easy. Maybe we just haven't learned yet where to look for the work.

Polina Kudelkina

Sources

• Yves Klein's Anthropometries and "living brushes" — MoMA, Guggenheim Bilbao
• Sol LeWitt's wall drawings and instructions — National Gallery of Art, Artforum
• Lady Eastlake, "Photography" (1857) — The Met, full essay archive
• "Bias against AI art can enhance perceptions of human creativity," Scientific Reports, 2965 participants across 6 experiments — Nature
• Millet et al., 4 studies, 1708 participants on AI art and perceived creativity/awe — cited in arXiv:2406.11844
• Art experts rating AI-generated work more harshly than non-experts — PMC study
• 432-participant study on authorship and copyright attitudes toward AI art — arXiv:2407.10546