AI-generated art is winning prizes, moving people to tears, and disrupting creative industries. The question of whether machines can truly be creative keeps getting harder to dismiss.
May 30, 2026·14 min read·AI & Society
In 2022, a painting generated by Mid journey won first place at the Colorado State Fair’s fine arts competition. The person who submitted it never picked up a brush. Half the internet celebrated; the other half was furious. I’ve been thinking about that reaction ever since, because the fury tells you something the celebration doesn’t: we’re not actually sure what creativity is, and we’re terrified of finding out.
For most of history, the assumption held pretty comfortably. Machines could beat us at chess. They could spot tumors on an X-ray faster than a radiologist. They could process payroll, translate documents, route packages. Fine. But creativity felt different. Creativity was ours.
That assumption is now under serious pressure. And the question it raises — can a machine truly be creative? — turns out to be a lot harder to answer than it sounds, mostly because we’ve never agreed on what creativity actually is.
Part One
Start here: we don’t have a definition of creativity that everyone agrees on. Poets, psychologists, and cognitive scientists have been working on this for a long time, and the honest summary is that it’s complicated.
What Do We Actually Mean by Creativity?
The psychologist Mihaly Csikszentmihalyi spent decades studying this. His answer was less romantic than most people expect: creativity isn’t some private spark inside a person’s head. It’s a conversation between the individual, the domain they’re working in, and the community that decides whether what they made actually matters. Under that definition, novelty alone isn’t enough. The work has to be recognized.
Which already creates a problem. Because if an AI produces a painting that critics call striking and original, and audiences genuinely respond to it — who gets to say it isn’t creative? Our intuition says no. But intuition isn’t an argument, and I’m not sure we can just leave it there.
Key Distinction
Researchers often separate two types of creativity: divergent thinking — the ability to generate many original ideas — and convergent thinking — the ability to refine those ideas toward the best solution. AI systems currently excel far more at the former than the latter, because evaluating quality requires judgment grounded in lived experience.
The philosopher Margaret Boden offers a more useful map. She breaks creativity into three types: combinational (mixing familiar things in fresh ways), exploratory (pushing further inside an existing style), and transformational (scrapping the rules entirely to build new ones). Current AI is genuinely impressive at the first two. The third one is a different story.
Transformational creativity is what produced jazz, cubism, and the theory of relativity. It’s not recombination. It’s the recognition that the whole existing framework is wrong — and the nerve to say so and build something else. No AI has done that. Not because it runs out of compute, but because you can’t statistically predict your way to a paradigm shift. By definition, the next one doesn’t live inside the training data.
Part Two
What AI Can Do (And It’s Remarkable)
Before we get too philosophical, it’s worth being honest about what AI can actually do right now. Because it’s genuinely astonishing, and dismissing it doesn’t help anyone think clearly.
Midjourney and DALL-E produce images with the kind of visual coherence that took human illustrators years of training to achieve. GPT-4 and its successors write poetry people cry at. Suno generates full songs — lyrics, melody, arrangement — in the time it takes to pour a coffee. These aren’t parlor tricks. The outputs are real, and people respond to them as if they are.
15B+AI-generated images created in 2024 alone, by some estimates
38% of creative professionals who now use AI tools regularly in their workflow
<2s average time for a state-of-the-art AI to generate a full-length song
How it works, at the bottom: large models train on enormous amounts of human-made content, learning statistical relationships between concepts, styles, and structures. Ask one to write a poem about grief in the style of Rumi, and it draws on everything it absorbed about grief, about poetry, about Rumi’s specific rhythms — and produces something that can feel startlingly original. It isn’t predicting the next word so much as navigating a space of vast learned possibility.
And in blind tests, people can’t reliably tell the difference. AI music versus human compositions — people guess wrong. AI paintings in galleries — people engage with them, argue about them, find them meaningful. By any external measure of what comes out, the case for AI creativity is surprisingly strong.
The Problem of the Training Data
Here’s the complication. Everything an AI produces is derived, at some level, from what humans already made. There’s no original spark. Every AI poem is a remix of poems humans wrote first. Every melody sits inside a space that human musicians defined.
The standard counterargument: humans aren’t so different. We also learn by absorbing what came before. Shakespeare lifted plots from Ovid. Picasso studied Velázquez obsessively before he started destroying the rules. Influence isn’t corruption — it’s how creative lineages work.
“Good artists borrow; great artists steal. But what happens when a machine steals from everyone, simultaneously, all at once?”— A question worth sitting with
But the difference isn’t the borrowing. It’s what the borrowing is for. When a human artist draws on an influence, she’s doing it to say something. There’s a message behind it, a perspective, an experience pressing to get out. Art is communication — one consciousness reaching toward another.
An AI has no message. No grief that makes it want to write about loss. No childhood that shaped what it finds beautiful. It produces outputs — often extraordinary outputs — because that’s what the optimization process produces. Not because it’s trying to be understood.
Part Three
The Consciousness Question
Which brings us to the thing nobody can sidestep: consciousness. Can creativity exist without experience?
When a novelist writes a character in pain, she’s drawing — whether she knows it or not — on every version of pain she’s actually felt. The specific weight of grief in the chest, not the stomach, where people who’ve never lost anyone expect it to be. The way a hospital waiting room has its own quality of silence. The particular shame of a childhood injury that wasn’t serious but somehow still stings. You can’t learn that from text. It’s not stored in words. It’s stored in the body.
AI doesn’t have a body. It’s never been cold. Never been embarrassed in a way it couldn’t shake. Never been bored on a long drive and had a thought arrive that changed something. Thomas Nagel’s old question — what is it like to be a bat? — was making this point: experience has a subjective quality that description can’t fully capture. The bat’s sonar isn’t just a fact about bats. There’s something it feels like to use it. And that something matters.
Is there anything it’s like to be an AI? Nobody actually knows. The question isn’t settled, and probably can’t be, not with the tools we have. But most people who work in this area think the answer is no — or at least not in any sense that resembles what we’re talking about when we talk about human experience.
Human Creativity
- Grounded in lived, embodied experience
- Driven by intention, desire, and meaning
- Shaped by mortality, emotion, and social belonging
- Can break entirely from prior frameworks
- Evaluated and refined through personal judgment
- Often slow, nonlinear, and inefficient
AI “Creativity”
- Derived entirely from prior human-made content
- Driven by optimization and probabilistic prediction
- Operates without self-awareness or stake in outcomes
- Extraordinarily skilled at remixing existing styles
- Evaluated by external human feedback loops
- Instantaneous, scalable, and infinitely patient
Though I’ll admit this starts to feel circular. We keep defining creativity in terms of the qualities humans happen to have, which conveniently guarantees the answer is humans. Maybe that’s the right move. Or maybe it’s motivated reasoning dressed up as philosophy.
Part Four
When AI Makes Something That Moves You
Here’s the thing that keeps complicating my thinking: AI-generated art genuinely moves people. Not all people, and not all the time. But sometimes, and the responses aren’t fake.
A musician I know described hearing an AI-composed piece and feeling the hairs on her arms go up. A poet friend read an AI-generated poem to his class and watched two students cry. These weren’t polite reactions. The art landed. It did what art is supposed to do.
Does it matter, in that moment, that a machine made it? The philosopher Denis Dutton argued that our experience of art is inseparable from our belief that a human was behind it — that there’s something he called “artistic performance” baked into appreciation itself. The knowledge that someone chose this word, this note, this brushstroke, and not another. Strip that knowledge away, and the experience shifts, even if the object in front of you is identical.
Dutton’s point suggests creativity might be partly something we attribute rather than something works possess. In which case whether we call AI output creative depends, at least in part, on the story we decide to tell about it. That’s either reassuring or deeply unsettling, depending on your temperament.
Research Finding
Studies have found that when people are told a piece of music was composed by AI, their reported emotional engagement drops significantly — even when the music is identical to a version they were told was human-made. Our experience of art is shaped not just by the work itself, but by our beliefs about its origin.
None of that is cynical, exactly. It’s just an acknowledgment that creative meaning doesn’t only live inside the work. It lives in the relationship between the work, whoever made it, and whoever’s experiencing it. AI disrupts that triangle, and we haven’t worked out what it means yet.
Part Five
What AI Is Actually Changing About Human Creativity
The more useful question, I think, isn’t whether AI is creative. It’s what AI is doing to human creativity. And the answer, so far, is complicated in both directions.
A lot of creative professionals say AI tools have genuinely freed them up. A graphic designer can generate ten rough visual directions in an afternoon instead of three, then bring her actual judgment to bear on the one that feels right. A novelist stuck on a scene can use an AI to throw rough material at the wall, then tear it apart and rebuild it in his own voice. A composer can hear a full orchestration of a melodic sketch in minutes, without spending years developing orchestration skills he doesn’t have yet. These aren’t trivial gains.
The piano analogy keeps coming up: AI is just a tool that extends what one person can express, the way the piano extended what a single musician could do. I find this partly convincing. But the piano didn’t make pianist training obsolete. AI might.
Because here’s the part that bothers me: the slow accumulation of craft — the failed attempts, the years of deliberate practice, the hard-won moments of breakthrough — that’s not just how you get better. It’s how you develop something to say. If AI shortcuts the process, we might end up with more output and less depth. More images. Fewer visions.
Then there’s the economic reality, which doesn’t have a philosophical consolation. Illustrators, voice actors, music composers, copywriters — these are real people watching AI tools replicate their work at a fraction of the cost. Whether AI is “truly” creative is an interesting question in a graduate seminar. Whether it’s gutting creative livelihoods is a different kind of question, and it’s happening now.
Part Six
The Future: Collaboration, Not Competition
The people doing the most interesting work with AI right now aren’t treating it as a replacement or a threat. They’re treating it as a strange new collaborator — one with no taste of its own but an inexhaustible willingness to try things.
Artists are using it to explore conceptual territory they couldn’t reach on their own. Musicians are feeding it their own work, studying what it does with it, using those transformations as a prompt for their next move. Writers are keeping it in conversation with their drafts, not to generate the final text but to be surprised by something they wouldn’t have thought of alone. It’s a weird relationship. It doesn’t map onto anything that existed before.
Some of the most compelling work coming out right now is explicitly hybrid — the human and the AI visibly tangled together, neither one fully in control. Artists who’ve leaned into that ambiguity rather than tried to resolve it are making things that feel genuinely new. Not AI-generated, not human-made. Something else.
“The question is not whether the machine can paint. The question is whether painting with the machine can reveal something about ourselves that we could not see alone.”— A framework worth carrying
There are still creative territories where AI runs up against something it can’t fake. Memoir, confession, testimony — forms that derive their power specifically from the fact that they’re true, and that a real person with something at stake is telling them. An AI can write in the style of a memoir. It can’t write an actual one. It has no stakes. Nothing it says can cost it anything.
And I think that might be what we’re really after, when we reach for creative work. Not just the beautiful object. Evidence that someone made it despite resistance — that it cost something, that the person on the other end struggled, and reached, and occasionally glimpsed something worth passing on.
Conclusion
So — Can Machines Truly Be Creative?
So: can machines truly be creative?
My honest answer is that the question is real, but the binary isn’t. If creativity means producing novel, surprising, high-quality work that other people find meaningful — AI clears that bar, and it cleared it faster than almost anyone expected. If creativity means the expression of a conscious perspective shaped by a life actually lived — AI doesn’t have that, and might never.
What I keep coming back to is that creativity was never a fixed property sitting inside individual works. It’s something that happens between a maker and an audience, mediated by everything both parties bring. AI has entered that space. It’s changed the shape of it. We’re still figuring out what that means.
What I do know is that the best human creators I’ve watched engage with these tools don’t seem diminished by them. They seem provoked. Irritated in productive ways. Pushed toward something they wouldn’t have reached alone.
The machine holds the pen well. What it can’t do is decide what’s worth saying. That part is still ours — at least for now.
Tagged:Artificial IntelligenceCreativityAI ArtTechnology & CulturePhilosophy of MindGenerative AI
A The Thoughtful Web
We write long-form essays at the intersection of technology, culture, and what it means to be human in a rapidly changing world. Every piece is researched, original, and written for people who want to think — not just be informed.
Continue Reading
The Empathy Gap: Why AI Cannot Replace Human Connection in Mental Health CareTechnologyWho Owns the Work? The Growing Legal Battle Over AI-Generated ArtCultureThe Last Analog Artists: Why Some Creators Are Deliberately Avoiding AIPhilosophyWhat Is Originality, Really? A Short History of Where Ideas Come From
