I’m Sick of Reading AI-Written Posts
They all sound like the same guy
I’ve been very active on Twitter — now X — and LinkedIn for well over a decade.
Not casually. A lot.
I read those platforms like a power user. I skim posts all day for interesting tools, small projects, weird experiments, detection ideas, and things my research team may need to look at, test, or write signatures for. A big part of my job is keeping an eye on what people build and what starts showing up out there.
So this is not coming from someone who reads five posts a week and suddenly has a theory.
This is coming from someone who has read an absurd amount of them.
And I really can’t stand this style anymore.
It feels repetitive, synthetic, overperformed, and weirdly smug at the same time.
Lately a huge part of my feed has started to sound like it was written by the same person. Not a real person, obviously. More like a machine that was trained on startup threads, product launches, fake thought leadership, and smug little “gotcha” posts.
The annoying part is that these posts are often not badly written.
They are polished. Clean. Nicely structured — and still unbearable.
The first tell: fake contrast everywhere
This is probably the most obvious one.
- “This is not opinion. This is math.”
- “It’s not a feature. It’s a paradigm shift.”
- “Not hype. Reality.”
- “This isn’t a bug. It’s the system working as designed.”
Same trick, every time. The post sets up two sides, then slams one of them down to make the other sound smarter, harder, more final.
It is cheap rhetoric and AI uses it constantly.
The second tell: absurd certainty
AI-written posts love absolute language.
Always. Never. Everyone. No one. Proven. Solved. Permanent.
You see it in lines like:
- “This changes everything.”
- “The old way is dead.”
- “We now know, with certainty, that…”
Real people usually leave a bit of space. They qualify things. They hesitate sometimes. They sound like they know the topic is bigger than one neat sentence.
AI does the opposite. It tries to sound final even when it is just rearranging probability.
The third tell: the rhythm is too perfect
A lot of these posts are built from the same Lego bricks:
short hook big claim three neat support points dramatic conclusion
And then those little sentence groups:
- “No vendor. No black box. No negotiation.”
- “They don’t pause. They don’t reflect. They guess.”
Once you notice it, you can’t unsee it. The text stops sounding human and starts sounding assembled.
The fourth tell: it is desperate to sound important
This is the part that really gets on my nerves.
Every post wants to sound like a historic moment. A release note becomes a manifesto. A benchmark result becomes a civilizational warning. A half-baked side project gets presented like the industrial revolution just happened on GitHub.
You get lines like:
- “This is the moment everything changed.”
- “What happens next will redefine the industry.”
- “We are witnessing a fundamental shift in how humans create.”
Everything is framed as if the reader is supposed to stop breathing for a second.
The real problem
The issue is not that AI text is easy to spot because it is bad. The issue is that it is easy to spot because it keeps using the same style markers over and over again.
A real person might write a sentence that is slightly clumsy but clearly theirs. An AI-shaped post usually feels smoother and emptier at the same time. It wants to sound sharp more than it wants to say something.
What’s missing is the part that makes writing feel like it came from someone.
What I think will happen
Something else is starting to happen though.
Once you’ve seen enough of these posts you start recognizing the tone immediately. The structure, the contrast tricks, the overly clean confidence. After a while you don’t even read them properly anymore. Your brain just flags them as “probably AI” and moves on.
And that changes how you value them.
Not because they are necessarily wrong. Many of them are technically fine. But the moment it feels generated you automatically apply a discount to it. I think of it as an authenticity penalty. The content might be okay, but you know the cost of producing it was basically nothing. A few prompt lines and a model doing the rest.
So the reader ends up investing more attention than the writer invested effort. That imbalance is noticeable.
There is also another thing building up which I would describe as synthetic fatigue.
People spend more and more time talking to generated voices. Support chats that clearly start with an AI agent. Replies that sound very polite, very structured, very careful — but also strangely empty. You can almost feel the system trying to keep you in the loop of friendly explanations instead of actually solving the problem.
After a while that just becomes irritating.
You want a real answer from a real person. Even if it is shorter, rougher, maybe a bit blunt.
What’s funny is that the actual quality of the text is not really the point anymore. The issue is the feeling that nobody actually sat down and wrote it.
There is a Japanese idea, wabi-sabi, that beauty comes from imperfection. From small irregularities that show something was made by a human hand.
I have the feeling something similar will happen with online writing.
When everything can be generated instantly and perfectly, the imperfect thing suddenly becomes the signal that a real person is behind it.
And people start valuing that again.
This is not about using AI at all
I use AI myself.
For spelling. For grammar. Sometimes to clean up a draft.
That’s not the problem.
The problem starts when the text no longer sounds like a person who had something to say.
It starts sounding like the same off-the-shelf voice we now see everywhere — polished, confident, repetitive, and weirdly empty.
That may be the real tragedy of this moment.
We have tools that can help people write, and instead we use them to make everyone sound the same.
