This was my thought, when I saw the post Marc refers too: I am glad you can be so confident something is AI, because most of the time I can't be sure (outside of maybe a few obvious cases). Marc's thread makes good points beyond this one
Marc F. Bellemare1) Great strategy if you are so smart as to have a rate of false positives be near zero. For the rest of us, it remains difficult to detect AI writing with any good degree of confidence. Not even AI can do that; we're dealing with a technology that manages to outdo itself from month to month. 2/