WRITING IN THE AGE OF CHAT GPT, or
"JOHN HENRY WAS A COPY WRITIN' MAN..."

first published on LinkedIn, June 2025

Let’s kick this off with some candor: ChatGPT is amazing. Remember the first time you used it? Sure you do. It wasn’t that long ago. I’m pretty sure that moment is going to stay with me the way other paradigm shifting before/after moments have—the first time I played Pac-Man; the first time I heard hip-hop; the first time I got high. I didn’t know what the future would look like, but I was pretty sure that thing was going to be a part of it.

Since then, my main use of GPT has been as a kind of toy. Give me Quint’s monologue from Jaws,  about the USS Indianapolis, but as written by Vladimir Nabokov. Make me a picture of James Joyce and Karl Marx playing Catan in a seedy opium den. List five reasons Vanilla Ice was better than the Rolling Stones. And voila, there it is. It’s like magic, the greatest piece of parlor trickery ever programmed.

I hear you can use it for other stuff too, like aggregating data, planning itineraries, conducting deep-dive research, and undermining the entire US educational system. And it does it all so seamlessly! It talks to you in plain language you can understand, leavened by a slightly jaunty, ingratiating tone, like an excited college tour guide or a fancy hotel concierge. For all of its relative facility with data and image, it’s the linguistic aspect that dazzles the most fully. We knew Google could uncover the most obscure details and references for any subject of your choice, but the way GPT shuffles that information into conversation-quality sentences is the killer app's killer app.

And this killer app has no shortage of victims. (Call GPT the serial killer app if you're so inclined.)  Everything from established teaching models to journalistic ethics to my next freelance writing gig is supposedly in danger of finding GPT's cheerful, ever-solicitous shiv in its back.  Maybe some of that concern is valid.  In a different lifetime, I might have taught English at a high school or university, and these days, I feel a little bad for that version of myself.  But I'm not worried about this version of me, and here's why.

Back in 1999, a quartet of writer/technologists published a document called The Cluetrain Manifesto. Modeled on Luther's 95 Theses, the Manifesto was a set of propositions drawn from the observed collision of networked communities with corporate practice.  One of the Manifesto's central claims postulated that as customers and clients formed online communities fed by their own voices, the more impersonal and anonymous “official voice” used by businesses to communicate with those customers and clients would be seen as increasingly inauthentic and ineffective. Now that consumers and employees knew what “real” voices sounded like in the digital world, value would accrue to voices that sounded recognizably human rather than the distant org-speak that came from the top of most corporations.

I found that insight compelling in 1999, and it remains so for me today. People want to feel like they are talking to other people, not faceless bureaucracies. And of course, it's GPT's uncanny ability to mimic the cadences of normal human speech that would seem to represent the nightmare endpoint of the proposition, whereby that value of person-to-person exchange is stolen and subverted by a simulacrum on one end of the line.

But here's the thing—every day, GPT gets easier and easier to spot.  It's not just the app's fondness for em-dashes that trigger the AI detectors. (I expect the first sentence of this paragraph caused someone out there to suspect I exist in a server farm somewhere.)  It's the reliance on rhetorical questions, its relentlessly discursive approach to generating conclusions, above all that direct, upbeat, blithely enthusiastic voice that tips its digital hand.

As more and more people and organizations use GPT, its presence is going to become increasingly obvious.  And within the not-too-distant future, GPT is going to calcify into the impersonal “official voice” that the Cluetrain guys identified in 1999.  And we'll be back where we started, just with a corporate voice where it's always casual Friday and it occasionally calls you “buddy”.

When you hire a writer, what do you want that writer to do?  Maybe you just want garden variety editing and proofing.  In that case, don't be a doofus—go with the app.  Maybe you just want to share information for the public record, as clearly and inoffensively as possible.  Likewise, that kind of job is GPTastic.

But chances are, when you hire a writer, you're doing so because you want your words to have impact.  You want them to move your audience to emotion or action.  You want the piece to intrigue, or incense, or amuse, or frighten.  You want it to stand out, maybe even jump out.  Usually, it's not enough just to put the information out there; you've got to make it worth reading.

If you're looking for GPT to do that, you're out of luck (buddy). And the reason for that is simple: GPT's work—every bit of its dazzling parlor trickery—is based entirely on replicating the patterns it's learned.  And impact is generated by the disruption of patterns.

GPT only knows which words typically follow which other words.  That's the entire ballgame.  It cannot make a move it hasn't seen before.  Its prose cannot surprise you.  It has no inherent grasp of the impact generated when a piece of writing shifts from elevated, polysyllabic diction into blunt, common vernacular.  It has no inherent understanding of phonics or syllables, of what words sound like when they string together in your head.  GPT is a program, and programs only follow rules. And good writing—really good writing with impact, and color and (dare I say it) soul—comes from knowing when and how to bend and break the rules.

On top of which, GPT will simply make stuff up and embarrass you, but that's a topic for another day.

So no, I'm not worried about the threat that GPT poses to me as a professional writer.  In fact, as GPT becomes more and more ubiquitous, the value of writing that stands out from and above GPT's is only going to climb. The intelligence behind Chris Green Words will remain proudly wrapped in its wet, squishy, carbon-based trappings.

Now forgive me, I have to go see what this article sounds like in iambic pentameter.

Get in touch for your next project.

Contact