AI can replicate human-made art. Here’s why it can never replace it.
The context you need, when you need it
When news breaks, you need to understand what actually matters — and what to do about it. At Vox, our mission to help you make sense of the world has never been more vital. But we can’t do it on our own.
We rely on readers like you to fund our journalism. Will you support our work and become a Vox Member today?
AI can replicate human-made art. Here’s why it can never replace it.
We can learn a lot from what the camera did to painting.
As AI continues to encroach on every aspect of our lives, there is a persistent fear or hope, depending on your angle: AI will someday take over art. The internet is full of quizzes showing that most lay people cannot tell the difference between AI-generated art (digital pictures of paintings, prose) and the real thing. Multiple studies have shown that when people are shown AI-generated art and human-made art, but are not told which is which, they tend to prefer the AI-generated art, whether it be images, poetry, or prose.
Yet what’s striking is that despite this disparity, people still consistently say that human-made art is what they want.
In one study published in 2023, participants were shown a series of images, each randomly labeled “AI-made” or “human-made.” Participants rated the images they thought were machine made as worse than the images they thought had been created by a human artist — even when those were actually human-made.
A natural experiment in how difficult it can be for people to tell the difference between AI-generated art and human-made art occurred last month, when the prestigious Commonwealth Foundation awarded its short story prize to “The Serpent in the Grove,” which a bears some of the hallmarks of AI-generated prose. In a statement to New York magazine, the Commonwealth Foundation said that the prize committee does not use AI checkers, but that “all shortlisted writers have personally stated that no AI was used.”
The big “tell” for “Serpent in the Grove” was that it is riddled with metaphors that are rhythmic and evocative at first glance but fall apart when you try to figure out what they mean: “The girl smiled like sunrise over a sink”; “She had the kind of walking that made benches become men.” If art is about connecting with another human mind, we might say that “Serpent” fails if, when you read it, you find it almost impossible to tell what the mind behind that story is trying to say.
One conclusion you might draw here is that the widespread disdain for AI-generated art is empty snobbery. If human-made art were so much better, the argument goes, then people would be able to see a real difference.
This line of thinking relies on the belief that “good” art is something that many people find appealing, at least in a vacuum. At this point, AI has automated that generation fairly successfully. At some point, it may get even better at it.
But I don’t think those study participants were lying when they said they wanted human-made art, even if they couldn’t tell the difference. Even if we get to a future in which AI’s persistent glitches are ironed out, so that there are no more missing fingers and garbled sentences, and AI-generated images and music and poetry and prose and film are completely indistinguishable from the best a human can produce, even to highly trained experts — even then, I think people would still keep saying they would rather experience art made by humans. And........
