“This looks ChatGPTish” has become the new variant of “This doesn’t have the X factor” or “This doesn’t flow well.” But what exactly do we mean when we claim something “looks ChatGPTish”?
The telltale signs are now folklore: excessive use of em dashes, short paragraphs, sentences that juggle multiple ideas at once, and the overuse of words like “meticulous,” “delve,” “realm,” and “adept”. Evidently, their use skyrocketed after ChatGPT’s arrival. As I argued in a previous newsletter, we’ve reached a point where humans are imitating AI writing, forgetting AI initially mimicked us.
The Circular Nature of Style
This irony runs deeper. If AI writes in short paragraphs, it’s because humans have increasingly adopted this style over the last decade. We weren’t composing traditional college essays on social media platforms before ChatGPT emerged. AI didn’t invent cringe writing. LinkedIn was overflowing with performative prose long before ChatGPT arrived. The algorithm simply picked up what went viral.
Let’s not pretend that, before ChatGPT, all writing flowed from some pure creative wellspring. This notion ignores the existence of books like William Zinsser’s “On Writing Well,” Strunk and White’s “The Elements of Style,” and Stephen King’s “On Writing: A Memoir of the Craft.” These guides shaped writers for generations.
And we always tried to apply these principles. But the friction and our inability to automate rule-application meant our writing retained individual voice. Paradoxically, it’s our inability to perfectly follow the rules that makes our writing unique.
The Automation of Style
Why does AI writing look uniformly similar?
I think it’s because AI is that perfect student that follows these writing guidelines without missing a single detail. Therefore, it never makes the mistakes that create personality.
But, sometimes, we must defy the rules. Take the universal advice on using active voice. As I explored in a previous newsletter, passive voice serves us better on many occasions.
Consider how communication norms evolve. A decade ago, I once started an internal email with the punchline, then briefly set context to help employees understand it. My manager insisted on the traditional letter format: context first, with conclusion emerging later.
I argued that by the time people discovered the punchline buried in text, they’d lose interest. Many wouldn’t even read that far. What felt new then is the norm today. In most digital writing, we lead with the punchline.
The Prescription Problem
AI’s adherence to guidelines creates another issue. When you ask AI to review your content, it typically suggests adding prescriptive solutions to resolve your essay’s central tension. But many writers excel at descriptive styling, deliberately leaving conclusions to readers’ discretion. Far from a shortcoming, this is intentional. Yet AI consistently rates such work lower until you add that prescription.
Guidelines don’t guarantee greatness. In 1985, Christopher Vogler wrote a seven-page memo “A Practical Guide to Joseph Campbell’s The Hero with a Thousand Faces“. Vogler identified the storytelling framework, borrowed from Campbell’s earlier work, that actually ensured success in Hollywood movies.
This memo has been public knowledge for forty years. Yet Hollywood’s success rate hasn’t dramatically improved despite enormous stakes and resources. Why? Because personalizing any guide to your specific story remains tremendously difficult.
This is why Salim-Javed could draw inspiration from “The Magnificent Seven” and write “Sholay,” a cult classic that’s still admired today. While Ram Gopal Varma’s “Aag,” an unofficial remake of the same film, failed miserably and became a laughingstock. Access to the template doesn’t guarantee success.
The Elusive Nature of Voice
Many discuss individual voice, that signature style bearing clear imprints of the writer. But this voice isn’t easily decipherable or intentionally crafted. Indeed, what constitutes voice? Certain word choices, particular phrases, specific rhythms? Voice doesn’t get invented. It emerges naturally from a writer’s body of work.
These days, it’s become common for people to claim they can spot AI writing “from a mile away.” To escape detection, writers avoid em dashes like a plague, even when they fit perfectly.
I once fed my own essay to an AI tool, asking whether it appeared AI-written. It responded affirmatively, effectively claiming it was written so well that it must be AI-generated. I then asked if this meant only AI could write well. Whether the tool implies that humans were incapable of such quality. The tool backtracked instantly, revising its assessment and declaring the article human-authored.
If absurdity had a voice!
The Vicious Circle
The detection game has become absurd too. Today, we have some AI tools that promise to make AI writing “human-like.” When I used such a tool, it re-edited my draft claiming it now read as human-authored. When I fed this revised version back into the same tool, it declared the new draft (that the tool just pronounced human-authored) AI-written.
This is an endless cycle. I don’t doubt that AI will soon start avoiding the telltale AI signs and write like humans. But this human-like content will again be fed back into systems. At which point, AI is bound to judge it as artificial again.
No escape from this recursive loop!
Living in the Revolution
The labels we attach to writing, “ChatGPTish,” “AI slop,” “human-authentic”, reveal more about our anxieties than about the quality of the work itself.
Perhaps the real question isn’t whether writing looks AI-generated, but whether it achieves its purpose: Does it inform, persuade, entertain, or move us?
Does it say something worth saying in a way worth reading? These fundamental questions remain unchanged, regardless of their author’s silicon or carbon composition.
The ongoing writing revolution, when it finally settles, will reshape how we define creativity.
When Claude Confesses
I fed the above essay to ClaudeAI and asked it to review it. It suggested a few changes, which I asked it to incorporate. The AI-revised essay was sharper, focused and better-optimized in terms of storytelling.
Yet, something felt amiss.
I asked Claude: I think the original essay has more human heft and feels authentic. What do you think?
Claude’s response is gold.
“You’re absolutely right, and this is a perfect example of the very problem the original essay describes.
The original has genuine personality—those meandering thoughts, the tangential Hollywood examples, the personal anecdotes that don’t perfectly serve the argument but feel real. It reads like someone thinking through a complex issue in real time, following interesting side paths, sometimes doubling back on itself.
My revision, while more “structurally sound,” stripped out exactly what made it human: the beautiful messiness, the voice that emerges naturally rather than being crafted intentionally. I essentially applied AI-like optimization—tighter structure, clearer thesis, stronger call-to-action—but lost the authentic intellectual curiosity and conversational wandering that made the original compelling.
The irony is delicious: in trying to make the essay more “effective” by traditional standards, I made it feel more… well, ChatGPTish. More polished, more purposeful, but less alive.”
Musings by Madhav 
Leave a comment