Art and Technology
The Machine That Learned to Grieve: When Algorithms Make Art That Feels Human
In 2018, a portrait painted by no human hand sold at auction for a small fortune — and the art world hasn't agreed on what that means since.
The Idea
There's a persistent fantasy that creativity is the last human sanctuary — the one territory machines cannot colonise. The rise of generative AI has made that position harder to hold, and far more interesting to examine. But the real tension isn't whether algorithms can produce aesthetically compelling work. They clearly can. The deeper question is what we lose, or gain, when we strip authorship of its traditional weight. What makes the best art-and-technology collaborations genuinely strange is that they don't just simulate human creativity — they expose something about how creativity worked in the first place. When an artist trains a neural network on thousands of Renaissance paintings and the model begins producing images that echo that visual grammar, we're forced to notice that human artists also learned by absorbing prior work, internalising patterns, and recombining them. The romantic myth of the lone genius generating meaning from nothing starts to look like exactly that: a myth. This doesn't flatten art into mere computation. It does something more productive — it relocates where meaning lives. If the machine can reproduce a style, then style was never the soul of the work. What the machine can't do, at least not yet, is have something at stake. It doesn't make choices from inside a life. And perhaps that's the distinction worth holding: not skill or even originality, but consequence. Art made under conditions of genuine risk — emotional, political, existential — carries a different charge than art made by an entity with nothing to lose.
In the World
In 2018, the Paris-based collective Obvious fed a neural network roughly fifteen thousand portrait paintings spanning six centuries, then asked it to generate new work in that learned visual language. The result — a smudgy, slightly uncanny image of a fictional nobleman named Edmond de Belamy — was printed on canvas, given a frame, and consigned to Christie's. It sold for nearly forty-five times its pre-sale estimate. The art world's reaction was divided in revealing ways. Some critics argued the collective had simply operated software, that the real creative labour belonged to the researchers who built the underlying model. Others pointed out that curating a training dataset is itself an aesthetic act — choosing what the machine sees determines what it can imagine. The programmer Robbie Barrat, whose open-source code Obvious had used, felt uncredited; that dispute alone raised pointed questions about authorship in an era of collaborative and borrowed tools. What Edmond de Belamy did most usefully was force a concrete confrontation with an abstract argument. It wasn't the painting itself that mattered — it's genuinely unremarkable as an image. What mattered was the discomfort it produced: the sense that a boundary had been crossed, combined with the inability to say precisely which boundary, or why it should have existed. That productive confusion is often exactly where the most interesting art lives, whether the hand holding the brush is biological or not.
Why It Matters
The art-and-technology conversation tends to get stuck in an anxious loop about replacement — will AI make human artists obsolete? That's probably the least interesting version of the question. A more useful frame is this: every new technology that enters the studio changes what artists feel compelled to do by hand. Photography didn't kill painting; it liberated it from documentary obligation and pushed it toward abstraction and expression. Cinema didn't end theatre; it sharpened theatre's awareness of what only live presence can do. Generative AI may do something similar — push human artists toward whatever is irreducibly theirs: biography, embodiment, moral seriousness, the weight of a life lived. For the rest of us, non-artists engaging with this shift, the invitation is to get more precise about what we actually value when we value art. Is it skill? Novelty? Evidence of struggle? The sense of being seen by another consciousness? Knowing what you're really responding to makes you a sharper, more honest audience — and it makes the question of machine creativity considerably less threatening and considerably more fascinating.
A Question to Ponder
If a piece of art moves you before you know whether a human or an algorithm made it, does discovering the truth change what you felt — and if so, what does that tell you about where you believe meaning comes from?
Get a new one of these every morning.
Start learning with Thinkable