An experienced copywriter, having finished a longer text, decides to just for laughs check it in “AI” detector software. It comes back saying 61% is likely AI-generated. For his own text.
But.
Wait for it.
The “AI” detector then offers a service…
“Would you like to humanize your text?”
…
The implication of this hurts my brain.
The only thing that realistically should happen if a generative system edits your text is that any estimation of it being “AI”-generated should increase. But the “AI” is now offering to reconstruct human text so that will appear less “AI”, and more “human”, to “AI” detection.
I’m out.