That's true. I notice this usually happens when the text length is too short for the detector to make a proper analysis. The detectors are not always reliable but if 3 detectors all say the same thing, there's a good chance its on to something. If 3 detectors say the same thing about multiple posts, there's an even better chance its on to something. Proclaiming a post to be written by AI because it followed the instructions in the quoted post... I'm not convinced that's an "official" method of detecting AI.
Spammers are very adaptive for spamming.
It's ok with them if one-lined or two-lined posts are considered as spam, they will be adaptive by making lengthy word-salad posts for non sense discussions.
Then, if we argue them about useless shit word salad posts, they will try to use Ai for achieving lengthy and seems to be informative posts.
If we consider lengthy and informative posts as one of signals of AI-generated posts, they will try to be adaptive again by using AI for post creation, but only trim lengthy given posts by AI to much shorter, very short posts for bypassing AI detective tools.
They are adaptive but with whatever shitpost approaches, their posts are always useless!
Before AI era, they even tried to use lengthy (but non sense) posts for merit exchange, as they think it is more accepted for lengthy posts with received merit.
