Types of AI phrasing that I hate (and that make it easy to spot AI-generated text):
1. "Who does what", "This or that", "Four things to do" - what, this, that, those, who, what; this really gets on my nerves because it sounds clumsy and foolish. It is at its worst when you see it on another page or slide and have completely lost context.
2. "Same thing", "Same that" - AI loves to repeat the word "same" a lot. Same what? If I skim or move to another page, I have no idea what is being discussed.
3. Pointless negations, such as "This, not that" or "It's light, not dark", "One feature, not many" - no shit, Sherlock?
4. Overly dramatic "movie trailer" language, like "ONE MAN, ONE GUN, ONE SURVIVAL", I see this a lot in official texts.
5. Excessive corporate jargon
@rolle
I think too that many texts are clearly LLM blaablaa, but still, just for shits and giggles, I've put some of my jobs place's managers' announcements etc. (without any sensitive details), to different AI detectors, and they are always getting 100% LLM grading.