Why Everything Feels Like It Was Made by AI

6 October 2026 · 4 min read · Syed Omar Faruk Towaha

Open a landing page, a LinkedIn post, a product description, a stock-looking illustration, a "personal" email from a company. Somewhere between the first and second sentence a small voice says: a machine wrote this. Sometimes it's right. Sometimes it's a human who has simply absorbed the machine's habits. Either way the feeling is now everywhere, and it's worth asking why.

My short answer: AI doesn't have a style, it has an average. The longer answer is below.

1. Models are trained to be the most likely thing

A language model predicts what usually comes next. An image model predicts what usually looks right. Push that machinery out at scale and you get the centre of everything it has seen: the median sentence, the median gradient, the median hero image. Nothing is wrong with the result, and that's the problem. Nothing is anyone's, either.

Human work has bumps: an odd word choice, a joke that only lands for people who know the author, a layout that's slightly too bold. Those bumps are exactly what gets sanded away when you optimise for "likely".

2. The tells in text

Once you notice them, you can't stop:

The rule of three. "Fast, reliable, and secure." "Clear, concise, and compelling." Three beats feels complete, so it appears everywhere.
"It's not X, it's Y." A fake contrast to sound insightful.
The tidy opener and the tidy closer. "In today's fast-moving world…" at the top, "In conclusion…" at the bottom, and a summary of what you just read in between.
Safe, borrowed vocabulary. Delve, tapestry, landscape, leverage, seamless, robust, unlock, navigate. Nobody talks like this at a kitchen table.
Em-dashes and perfect rhythm. Every sentence the same length, every paragraph the same shape.
Zero specifics. No names, no numbers, no "last Tuesday", no mistakes. It talks about everything and has been nowhere.
Eager-to-please tone. Never grumpy, never unsure, never rude. Real people are all three before lunch.

3. The tells in images and design

The same averaging shows up visually: the glossy, over-smooth skin; the perfectly centred subject with cinematic teal-and-orange light; hands and text that are almost right; the purple-to-blue gradient behind a rounded card with a soft shadow. Generated sites converge on the same hero section, the same three feature cards, the same pill buttons. When ten thousand pages are built from the same defaults, they look like siblings.

4. We are also training ourselves

Here's the part I find more interesting. Plenty of "AI-looking" content is written by people who spent a year reading AI output and now write like it. Templates, engagement-bait formats and "10 lessons I learned" threads were already flattening us before the models arrived. The machines learned from us, then we started learning from them. It's a feedback loop, and the loop pulls towards the middle.

5. Why it feels bad, not just different

Effort is a signal. A handmade thing says someone cared enough to spend time on me. Free, infinite text says nothing about care, so I stop extending trust.
Sameness is boring. Our brains are pattern detectors; the tenth identical pattern is just noise.
Suspicion costs attention. Half my brain is now asking "who made this?" instead of "is this true?" That's a real tax on reading.

6. What I do about it in my own writing

I'm not going to pretend I never use these tools, so here is the practical bit. The fix isn't hiding the tool, it's putting yourself back in:

Start from something only you have: a real number, a real failure, a real opinion you could be wrong about.
Cut the symmetry. If three examples are listed, keep the best one. Let a short sentence sit next to a long one.
Swap the borrowed words for the ones you'd actually say out loud.
Keep one rough edge on purpose: a joke, a tangent, an unfashionable view.
Read it aloud. If it sounds like a press release, it is one.

The bottom line

"Looks like AI" is mostly shorthand for looks like no one. The cure isn't avoiding the technology; it's remembering that the valuable part of any piece of work was never the polish. It was the person behind it, with their specific history, taste and mistakes. Polish is now free. Specificity is what's left.