AI Doesn’t Write Better Than You. It Just Writes What’s Safe. And That’s the Problem.

I spent an hour reading two short stories. One was written by a human, the other by an AI. I couldn’t tell which was which. And when I found out, I was ashamed to admit I liked the AI one better.

You’ve probably seen the headlines: “Study: People prefer stories written by AI”. But the real story isn’t that AI wins — it’s that we’re losing our taste for anything that isn’t polished, predictable, and safe.

The real risk isn’t that AI replaces great writers. It’s that AI will make great writing feel wrong.

Here’s what the study actually measured: 1,500 participants read two stories — one human, one AI-generated — and most couldn’t tell the difference. Worse, when asked which they liked better, they leaned toward the AI version. The researchers called it a “preference for the average.”

But let’s call it what it is: a quiet catastrophe. The AI didn’t write something brilliant. It wrote something smooth. It averaged out every human voice that had ever told a similar story, and that average felt comforting. Familiar. Like a warm, vanilla-flavored lie.

One comment on the study nailed it: “People also like superhero films, but they are not exactly lauded as the pinnacle of cinema.” Exactly. Broad appeal isn’t the same as quality. But here’s the twist: as we consume more AI-generated content, our brains will start expecting that smooth, average texture. The jagged edges of real human experience — the messy, the weird, the raw — will begin to feel like noise.

We are not just teaching machines to write like us. We are teaching ourselves to write like them.

Think about the last time you read something that made you uncomfortable. Not because it was offensive, but because it was different. A story that didn’t follow the beat. A sentence that didn’t resolve into a tidy insight. If you’re like most people, you probably scrolled away. That’s the AI training in action.

I’ve seen this firsthand. A writer friend submitted a piece to a publication. The editor said it was “too original” — confusing, risky. They asked for a rewrite. The writer refused. The editor then ran the original through an AI, got a “cleaner” version, and published that. It got more clicks. But it was hollow. The writer’s voice was gone, replaced by a statistically perfect ghost.

This is the future we’re sleepwalking into. Not a world where AI generates masterpieces, but a world where the average is the ceiling. Where the most successful content is the most predictable. Where the reader’s comfort zone shrinks until it’s a closet of safe, familiar stories.

If you can’t tell the difference between a human and a machine, the machine didn’t get better. You stopped paying attention.

So what do we do? First, stop pretending this is a neutral tool. It’s not. Every AI-generated story that gets published is a vote for the average. Second, start reading like a critic. Ask yourself: Did this feel too easy? Did it resolve too neatly? Did I learn something about the author’s life, or just about the algorithm’s training data?

And third — write. Not for clicks. Not for the algorithm. Write for the one person who still wants to feel something real. Write the weird sentence. The uncomfortable ending. The story that might confuse a machine. Because the machine is already trying to make you forget what that feels like.

I’ll end with something I wrote myself, not an AI. It’s imperfect. It’s jagged. But it’s mine:

“The most dangerous thing AI can do to us is not replace our jobs. It’s replace our taste. Once we stop recognizing the human, we’ll stop wanting it.”

And that’s a loss no algorithm can fix.

FAQ

Q: Doesn't this study just show that people prefer what's familiar, not that AI is better?

A: Exactly. The study measures broad appeal, not quality. AI optimizes for the average of human writing, so it feels smooth and predictable. That's great for clickbait, terrible for art. The problem is that as we consume more of this average content, we start to prefer it, and the genuinely original voices feel alien.

Q: What's the practical takeaway for a writer or content creator?

A: Stop trying to sound like an AI. The algorithm rewards safety, but your readers have a deeper hunger for something real. Write with specific details, messiness, and personal voice. And start reading critically — ask yourself if a piece felt too easy or too clean. If it did, it might be artificial.

Q: Isn't this just elitism? What's wrong with content that's broadly appealing?

A: Nothing, on its own. The danger is when the entire ecosystem — from publishers to platforms to reader expectations — is optimized for the average. That's a monoculture of taste. We lose the outliers, the experimenters, the ones who push us to think differently. If AI makes everything feel like a superhero movie, we'll never get another <em>Eraserhead</em>.

📎 Source: View Source