Artificial Intelligence
The Brilliant Illusion of AI Cognitive Theater
Is writing dying a slow death of at the hands of LLMs?
Posted April 21, 2025 Reviewed by Michelle Quirk
Key points
- Writing is being replaced by performance that only sounds intelligent.
- We confuse sounding smart with being right as real thinking fades from the stage.
- AI writes with style but often without substance, and thinking still starts with a pencil.
Pick up that pencil. There's an important lesson here, and I'm writing it word for word, leveraging my word prediction machine, my neocortex. As I see it, we’re living in an era of language that performs intelligence. Not just mimics it, but performs it.
You’ve probably read one of these stories or posts. It opens with a grand idea—coherence, curiosity, quantum cognition. It speaks of entanglement and epistemic mass, of conceptual wormholes and phase alignment. It weaves physics with philosophy and wraps it in a layer of shimmering metaphors. It's a full plate of cognitive mush that's served up with the appropriate garnishes of validation. Sometimes, it even seems sorta brilliant.
But more commonly, at least to me, something’s off. You reach the end of the pontification and realize you’re not quite sure what it said. Or worse, you feel like it said everything, but proved nothing. That’s the moment when the illusion breaks and falls into the dust of ambiguity or even irrelevance.
This is the new face of writing in the age of AI. It's a kind of cognitive theater where fluency masquerades as depth. And it’s only going to become harder to tell the difference.
When Language Gets Too Good
Large language models (LLMs), by design, are masters of mimicry. They’ve consumed the world’s libraries and learned how ideas sound when they’re smart. They understand rhythm, emphasis, and cadence. They can draft a manifesto on epistemology that reads like a blend of Borges, Bohm, and Baudrillard—with a hint of Nosta for that essential level of pander. Of course, that fluency is deceptive.
Because behind the elegance, there’s often no grounding. No meaningful cognitive flow or architecture that can be interrogated. These aren’t ideas—they’re performances of ideas. And when the performance is good enough, we applaud even when there's nothing on the stage. And perhaps in today's world, it's no longer the emperor but the author who is wearing no clothes.
It’s not that these texts are wrong. It’s that they’re untethered. They’re simulations of insight, optimized not for truth, but for the sensation of understanding. And I’m left wondering, when did it become better to sound smart than to be smart?
Ridiculously Recursive and the Fractal Trap
There’s a particular style that’s gaining traction. It's recursive, self-referential, heavy on systems that mirror themselves, and amplifies the prompter—however ridiculous it may be. And it all sounds curiously profound. But depth is easy to fake when you don’t need to explain the mechanics and you’re not accountable to empirical scrutiny.
This is what I call the fractal trap. It's the seduction of structure that looks intelligent because it echoes how we think. It builds on recursion, symmetry, and complexity. But complexity alone isn’t a sign of cognition. It’s often a smokescreen.
We’re wired to find patterns, to seek coherence. So when we see language that folds back on itself, when we read phrases like “phase-locked epistemic alignment,” our minds light up. Not because we understand, but because we’ve been conditioned to respond to the appearance of insight.
Bread and Circus for the Mind
This isn’t an academic concern; it's a cognitive crisis. Because as these systems evolve, the difference between real thought and synthetic thought-performance becomes harder to detect. And the risk isn’t that we believe the wrong thing. The risk is that we stop asking whether belief is warranted at all.
In a world where everything sounds smart, skepticism becomes the first casualty. We start to accept this perceived coherence as proof. We nod along to texts that dazzle but don’t anchor. And we quietly lose our ability to distinguish between language that thinks and language that fakes thinking.
Using an LLM Doesn’t Make You Less of a Thinker
Here’s the irony. I use an LLM all the time. Not to write for me—but to think with me. It’s a cognitive tool, a thought partner, a way to accelerate ideas and challenge assumptions. But I never mistake it for an oracle. If fact, I find the output increasingly derivative and rather vapid.
LLMs are the smartest and most stupid editors I've ever worked with. But I still claim authorship.
Pencil in hand, that job is still mine.
Closing the Curtain
This brilliant illusion is seductive. It flatters our intelligence, and, perhaps most dangerously, it wears the mask of insight. But we need to develop new habits and instincts that separate real thought from its simulation.
Remember, not everything that sounds brilliant is.
Today, discernment may be the most important skill we have left. So, do yourself and humanity a favor and pick up that sharpened pencil.
