detecting AI content 2 min read

i often get the odd feeling when i read other people’s posts,
that some posts are somehow AI generated.

because the text seems just too perfect?
copywriting-blocks-structured? too polished?

I don’t know.

it feels like they’re AI-generated, but I just can’t prove it.
I only get this slight, subtle annoyed feeling.

but then, I start to wonder:
what if I copy pasted that post content into AI itself?

can AI tell if a post is AI generated?

and so I did.

and apparently:
yes, AI can determine which posts are from AI, and which posts are not.

I pasted people’s content into AI

AI said, they are AI generated. like 85-90% AI.

too well structured, grammatically flawless,
almost perfect rhythm and pacing,
play in the safe area avoid the extremes, controversy, or truly strong opinion,
missing individuality/self-awareness, and mostly too generic
and there is one unique thing that almost only AI use:

AI will use the special quotes (“”) instead of the ordinary “” which humans often use.
and of course — the infamous emdash.

my own content

then I pasted my posts into AI.
chatgpt said that my post is 90-95% human.

the post is not well structured (ouch! 😂)
not grammatically correct (ouch! 😂)
some typos here and there, missing letters,
there are emotional inconsistencies (ouch! 😂)
and so on.

apparently, it is the imperfection that makes us human.
AI is just too perfect; too good.

so, now I can sleep slightly better,
because I already know which users are using AI and which users are not,
I don’t have to be suspicious all the time ever again.