0
I’ve seen polished human writing flagged as AI-generated, while lightly edited AI text sometimes passes as human. That makes me unsure how much confidence anyone should place in detector scores, especially when the writing has been revised several times.
Have you found these tools useful for anything beyond a rough signal? I’m interested in how teachers, editors, or publishers interpret a result without treating it as proof about who wrote a piece.
