Words, human- or robot-generated, cannot describe how tired I am of the Surefire AI Tip-Off scavenger hunt. It’s as tedious as, say, the “It’s not [X] — it’s [Y]” construction in an AI-written blog post.
This isn’t me coming out in favor of the widespread acceptance of AI-generated work, or saying the growth of AI use doesn’t matter — it matters. It has consequences. To the creative industry, to the economy, to the labor market, to our political system, to society, it matters and has consequences, and yet another LinkedIn post with hints on how to tell if your writer is trying to sneak robot content past you is helping with none of it.
The employers who’ve said an article read just fine, but they put it through an AI checker anyway and it came up 64%? The writers who’ve put their own work through checkers — totally non-AI-written, pre-ChatGPT even, work — and had it come up 80% robot-written? The writers who’ve talked about excising em dashes from their writing, their perfectly fine writing, for fear of being accused of AI tomfoolery?
Please: Stop with the lists. Stop with the checkers. Stop with the LinkedIn posts that… Eh, just stop with the LinkedIn posts. Don’t stop caring about the implications of Ai usage, but stop trying to clock AI one dead giveaway at a time. Just clock bad writing.
Just… clock bad writing.
Maybe it’s AI, or maybe it’s just bad writing.
Remember that all bad AI writing started as a computer remix of an entire internet’s worth of amassed good writing.
“The Rule of Threes” is good writing. It’s nice and balanced, it’s rhythmic, and it satisfies our very human love for a nice pattern. But if it’s used too often, it feels repetitive. It stands out. And some AIs do have this specific yet indescribable way of using it that’s just, like, “Yeah, that’s a robot.” But human- or AI-generated, repetitive phrasing and contorted language are bad writing.
“It’s not [X] — it’s [Y]” is a fine structure for drawing comparisons and contrasts. “You might think it’s this thing, but BOOM! It’s this other thing!” It’s got nice parallelism that makes it very satisfying to read. Used excessively, to the point that it becomes noticeable, it’s unsatisfying. Whatever writer, of the human or robot variety, put it on the page needed to dial it back, because that much not-X-but-Y is just bad writing.
“But here’s the thing:”, “And honestly?”, “The result?”, all decent transition markers. They can bring you into the final big idea, or highlight a major conceptual turn, and that’s good. Used too much, they make the writing sound choppy and add a performative kind of “just folks”iness to it. And if the writing feels choppy and hey-guys-I’m-just-like-you, a robot could be at fault, but if the writing feels and sounds fine but there’s a “Then I realized:” in there… maybe the writer just realized something.
Underscore, leverage, amplify, and seamless are words in common usage. Like, by humans. (Delve and tapestry, probably less so.) Just not all humans in all circumstances. So if your article about mini golf is talking about leveraging pivotal insights for defeating the Goofs the Clown hole, that phrase might could be out of place, no matter who or what put it on the page. It could be worth your time to give the copy an extra once-over to see if anything else potentially robotic jumps out. But J’accuse!ing your writer of Fraude over what could just be poor word choice is excessive.
I’m not even going to dive back into the em dash thing. I just can’t.
Man vs. machine
It’s dehumanizing. In a quite literal way. It’s taking a person’s work and scrutinizing it for hints they might be less than human.
One of the reasons I do value real, genuine, human-generated text over the AI version is that the real version is the only one that comes from thought. It’s the only one that involves human insight, experience, considerations, all coalescing into coherent ideas. AI-generated text comes purely from calculation — it’s statistical prediction of what word should go after another, based on an analysis of things actual, real, thinking people have already said. This instant suspicion at the first sign of the first em dash isn’t just an accusation that the person’s too lazy to write — it’s that they’re not human enough to think.
Repetitive writing. Tedious writing. Writing that lacks perspective. Language that doesn’t connect with the audience. A tone that’s too smooth, perfect, polished to fit the brand voice. Structure that feels more rehearsed than edited. Writing that’s more strung-together phrases than ideas. That whole uncanny-valley thing that’s common but not exclusive to AI writing — it’s bad writing, however you come by it, however it’s produced, it’s just bad.
Bad writing is bad writing, and during this Era of AI, it’s reasonable to include “was it a robot” in your assessment of why the writing is bad. But running to an AI checker at the first sign of a tricolon perfectly at home amid perfectly fine copy is… bad. You’re substituting an unreliable robot’s judgment for your own, you’re getting yourself into a lather over copy you would have given a gold star in pre-GPT years, and if the copy was, in fact, 100% organic, you’re the one dragging an AI into the conversation.
Plus, I am just sick to death of yet another article by yet another AI detective with the exact same list of clues. That is bad writing, too.
The fight to save humanity rages on.
I am, in general, very much opposed to telling other people what they should think or care about. Your mental state, your business. But… y’all.
The fact that the LLMs in common usage were trained on text yoinked from writers without permission or compensation? The fact that ChatGPT uses up to a water bottle’s worth of water per short conversation, the training process uses 120 homes’ annual worth of electricity, and the rapid spread of data centers is crowding out the housing market? The ability (and use) of AI-generated images to falsify the world around us and basically break our perception of reality? All worrisome. “Does that em dash mean this whole article was written by a robot?” Y’all.
And none of this is to say it doesn’t actually matter if something was written by a human or an AI — it does. It’s the difference between human cognition and computer calculation, human perspective vs. remixed insight, and that does make a difference. But these endless LinkedIn rants and tedious LLM scavenger hunts are only making the entire AI controversy and chaos that much uglier.
I’m not going to call it a witch hunt, mostly because witches didn’t actually exist and AI fluffery and fakery definitely do. It’s more a Cylon hunt — the Battlestar Galactica cyborgs hiding among us. Maybe they really are a sleeper agent, or maybe they’re just a shifty but ultimately human Aerilonian opportunist, and if we’re just going to go around spacing everyone we catch using an em dash, we’re not going to end up with better writing — we’re going to end up with a lot of good writers out the airlock, and probably some skinjobs left behind that write like robots but know better than to use an em dash.