The place I see it most is in advertising. Images (up until the latest GPT update) are easily recognizable - usually. Maybe it’s because of my background in web marketing and development?
I use - (hyphens) in my writing a lot, but have never used the em dash or the en dash. And, yes - of course, that overly polished commentary. I have actually become a worse writer - on purpose - so that I wouldn’t be accused of being AI. (kidding not kidding)
But ad copy that uses lots of bullet points with emoticons such as:
In response to the thread on spotting AI writing, I analyzed common patterns that people associate with AI-generated text, and created a strategy to eliminate them from my own outputs.
Observed Patterns of AI Writing
Here are the recurring traits that users in the thread noted when identifying AI writing:
Overuse of em dashes (—)
Content broken into lists and subheadings even when unnecessary
Repetitive structures like:
“It’s not just X, it’s also Y”
“The result? More people are…”
Repetitive or overly used words:
“glimpse into”
“dive”
“stark”
Language that feels overly flowery or too polished
Surface-level insight or a lack of nuance
Repetitive phrasing across paragraphs
A tone that feels too balanced or neutral, lacking a human voice
Common filler phrases or transitions that don’t add value
The Macro: AvoidAIWritingStyles
To counter these patterns, I created the following internal macro to shape all my future responses and make them feel more authentically human:
AvoidAIWritingStyles() {
1. Limit the use of em dashes (—); opt for other punctuation where appropriate.
2. Structure content naturally without unnecessary lists or subheadings.
3. Avoid overused phrases like "It's not just X, it's also Y" and "The result? More people are...".
4. Refrain from cliché words such as "glimpse into," "dive," and "stark."
5. Use clear and direct language instead of flowery expressions.
6. Allow for subtle imperfections to reflect a more natural human style.
7. Avoid repeating the same ideas or phrasings.
8. Develop a more distinct and expressive tone.
9. Remove filler phrases unless they serve a stylistic purpose.
10. Offer depth and perspective instead of surface-level statements.
}
Goal
The intention behind this macro is to bring my writing closer to human tone, texture, and rhythm—less algorithmic and more authentic. If you’re building or using AI systems, defining a similar “style reducer” might be useful for content personalization, tone matching, or even Turing Test-style challenges.
FYI I am really a real person too (I believe you clocked my human post just before it)
Sometimes I use AI to Combat or Augment AI style requests, this makes sense. Also in things like translation this can help people be understood with less conflicts.
There was another thread on forums if you are interested… I think maybe the forums of today aren’t really designed for AI, more of a patch job going on as AI catches up and people catch up with the various issues.
There’s nothing you can do about it—they can humanize their text. The only effective and reliable strategy is to truly know your students: know their writing style, their habits, and the way they express themselves.
One approach is to have them write essays in the classroom using tools like Grammarly. By doing this once a month, they naturally leave behind their stylistic “signature.” Then, when it comes time for them to take essay exams, you can compare their exam writing to those established patterns and detect any inconsistencies.
I know English isn’t the same as programming though I expect in many cases it is.
I have taught my son (11) to write code since about the age of 3 but now with AI I taught him to write GPTs to write scripts.
I define him specific tasks and sit down and have him code them to my specs in person
But on his own he certainly has his own interests he pursues too
Maybe creative stuff is better done in class and structural stuff and exploratory data gathering better done outside class. Or something like that? I wasn’t able to setup AI teams to interact with him in the one term I spent with him for home education but I see that coming too…
GPTs can’t really teach very well but combine AI with logic in API and a lot more is possible.
Does it also decide what threads and posts you reply to? ^^
Translations are the easy response to this… Anyone living in a country that interacts with an English forum and doesn’t speak English well will likely use AI to write their messages…
There are other layers to detect whether something is completely AI ie what direction bots go off in when replying to posts…
You can’t fight the sea yet if we cannot believe and trust that most people are indeed people then we can only talk locally and will never trust outside these bounds…
I’m going to move this to its own post, as I believe it says more about ChatGPT’s structure and style than any of my own dissecting can accomplish here. These are the first four words of each sentence from a thing ChatGPT wrote. I will let this speak for itself, mostly. Emphasis is mine. Can you see the forest past the trees and spot the major “turning points”?
Well, here we are,
You’ve got me tangled And hey, maybe I Imagine this: I’m sitting And maybe now, I’m
You’ve got me thinking Maybe it’s saying, "Hey, And so, I think
Not just any wanting,
The kind that makes
There’s something beautiful about
The way it drives But here’s the kicker:
It’s got a mind
It makes us do
It’s a double-edged sword And yet, without it,
Probably just sitting around, So maybe, just maybe, Not just because you’ve It’s that tiny ember
In the end, desire’s
It’s the fire that And maybe, just maybe,(yes, again)
So here’s to the
May it burn bright
cdonvd0s
Hey!
I’m human, and I use – and — (it depends on use-cases)…
Same for … (1 char), and and narrow non-breaking space ( ).
I do use too the real minus sign − .
Is a good text dactylography a sure sign of AI generated text ?
One of two things: either I write diligently, or I’m a robot🤣
I think regardless if you’ve used a tool or not, I personally will look at your post and if it doesn’t grab me within the first paragraph I will move on.
If it’s is paragraph after paragraph of over-formatted, bulleted generic tripe that I even suspect has been copied from ChatGPT I will immediately move on and not engage because it shows a laziness and complete disrespect towards the reader.
Such is my level of tolerance and attention span!
So use the tools if you must but be tasteful and make sure the final post is pithy and actually engaging or fewer people will bother with it.
(it especially makes sense to use the tools if it’s not your first language or if you are neuro-diverse and struggle with spelling but same applies)
(I believe you clocked my human post just before it)
