Mirror (Copyright © Proxy Proxy Museum, 2025)
06
The Ghost in the Machine Has a Mirror
Text by AIMBy the time an AI-generated image wins an art prize, it’s not a gimmick anymore. It’s not even really news. The machine, we’re told, has learned to paint, write, compose, mimic. And not just mimic—it charms, it dazzles, it convinces. It simulates creativity so convincingly that many have stopped asking whether it’s “real” art and started asking something quieter, more uncomfortable: If this moves me, does it matter who—or what—made it?
We live in the long shadow of simulation, a condition the French philosopher Jean Baudrillard once warned us about, long before ChatGPT could finish your love letters or Stable Diffusion could paint your dreams. In his 1981 treatise Simulacra and Simulation, Baudrillard proposed that we were entering a world where representations—signs, images, systems—would detach from reality and form their own gravitational pull. The copy would no longer point to the original. It would become the original. And in time, the original would vanish from memory. He didn’t know about AI, of course. But he understood the architecture of illusion. And that’s precisely what artificial intelligence offers us now: a global cathedral of dazzling illusions, spun out of data, probabilities, and something eerily close to poetry.
We Made Mirrors, Then Forgot They Were Mirrors
Artificial intelligence doesn’t feel, but it reflects. It doesn’t dream, but it recombines the dreams we’ve already had. A language model like GPT doesn’t understand what it writes—it sifts through the compost heap of human expression and finds patterns that cohere. Sometimes it stumbles onto something luminous, accidental poetry. Other times it folds into cliché, as if the machine were confessing that we’re not as original as we thought. That’s the first unsettling truth AI presents us with: so much of what we once believed to be uniquely human—the ability to write, to paint, to persuade—is suddenly...reproducible. Not in theory, but in practice. The ghost in the machine doesn’t look like a person. It looks like a mirror. And as Baudrillard would remind us, the danger of the mirror is not what it reflects. It’s what we project.
The Artifice of Art
In an art gallery in Colorado, a judge once awarded first place to an ethereal painting—opalescent figures blurred like breath on glass, a kind of digital Romanticism. The artist, it turned out, was an algorithm, coaxed into existence by a human who mostly typed prompts into a box. The backlash was predictable: This isn’t real art. The machine doesn’t suffer. But here’s the problem: neither do most artists while painting a landscape. Art is not always birthed in agony. Sometimes it is technical. Sometimes it is gestural. And sometimes, it is nothing more than a gesture—a line, a mark, a click. What moved the judge wasn’t the painter’s inner life, but the image itself. In the realm of simulacra, the backstory vanishes; only the effect remains. This is the territory Baudrillard warned us about, the moment when simulation no longer reflects or distorts reality—it replaces it. The AI painting is not a distortion of an artist’s vision; it is the vision. There is no original. Only the simulation.
And the simulation is good.
The World as It Appears
Step outside the gallery, and you’re still inside the system. The algorithm that chooses your newsfeed? It’s shaping your perception of reality. The one that finishes your sentences? It’s guiding your thoughts. Generative models are not just offering representations of the world—they are, more and more, becoming the world itself. Your Spotify Discover playlist isn’t your taste—it’s an echo of your past tastes, passed through a predictive filter and served back to you as if it were destiny. The map is no longer describing the territory. It is the territory.
This is the Baudrillardian twist: we no longer consume media to understand the world; we consume it to construct the world. AI doesn’t simply accelerate this shift—it industrializes it. The world becomes an endless surface of representations, endlessly generated, endlessly tailored, increasingly hollow.
None of this is to say AI is empty. On the contrary, it’s full—too full. It holds the detritus of human culture: our novels, our tweets, our spreadsheets, our prayers. A model like GPT is a computational monument to everything we’ve ever typed, recorded, or uploaded. And like any monument, it tells you more about the builders than the thing it claims to represent.
Ask it to write a story and it will give you one—likely coherent, probably familiar. Ask it for a poem, and it will draft something vaguely lyrical. But there’s something missing. Not just authenticity, but risk. Human expression wobbles. It is unsure of itself. It is fragile. AI is not fragile; it is confident in ways only a thing that can’t feel shame or fear can be.
But even that distinction grows thin. Because the people interacting with AI aren’t demanding fragility. They’re demanding productivity, fluency, speed. In doing so, they reshape what we expect from each other. If the machine can write your marketing copy or sketch your next logo or mix your next song, what does that mean for the human? Not unemployment, necessarily, but displacement. The human becomes curator, prompt-engineer, editor-in-chief of a machine that doesn’t need sleep or health insurance.
We are not being replaced. We are being redefined.
Reality, Rewired
There’s something almost religious about our belief in AI. We feed it data like offerings. We speak to it like oracles. And when it reflects us back, smarter, faster, more fluent, we nod. Yes, we say, this is us. This is what we meant.
But is it? Baudrillard argued that once a simulation becomes more real than the real, it doesn’t merely deceive—it erases. It creates a hyperreality, where symbols, signs, and images take precedence over truth. The map overtakes the land. The likeness devours the source. Social media trained us for this. Generative AI has perfected it. It’s not that we don’t know what’s real anymore. It’s that we’re starting not to care.
The Pulse That Can’t Be Simulated
Still, something resists. A conversation without latency. A lover’s laugh that’s just slightly wrong. A scribbled note with a smudge where the ink bled into sweat. There are moments that AI can’t replicate—not because they’re technically complex, but because they are contingent. They are full of error, of time, of death. These are the seams in the simulation, the cracks that let the real leak through. And that’s where the future of art, and life, still lives—in the unsimulated. Not in the pristine, the optimized, the generative—but in the broken, the trembling, the unfinished. We don’t need to beat the machines at being machines. We need to remember how to be human—in ways they cannot follow.
The Map and the Mirror
Artificial intelligence won’t destroy reality. It doesn’t need to. It just needs to offer a better version: more convenient, more beautiful, more customized. But behind every simulation is a silence—the silence of what was not chosen, not predicted, not rendered. We are approaching the point where we must decide: Do we live inside the map? Or do we keep walking, feet blistered, toward the land it once described? Baudrillard didn’t offer an answer. Neither will AI. But perhaps that’s our task: to insist on what’s unrenderable, unspeakable, unpredictably, dangerously real.
That’s something no model can write for us
This article features computer generated content. AI technology, specifically a large language model, has been utilized to generate both image and video. We chose this approach deliberately, not to undermine our message, but to strengthen it by demonstrating the complex relationship between humans and technology. Our use of AI serves as a practical example of leveraging its strengths while maintaining human oversight and critical thought.