When looking at the context of a given text, use of certain words or punctuation, can very well indicate AI use.
The "original" example was delve. There is no doubt that AI (did, or still does) use this word at a significantly higher frequency than the average person. I would say the same about em dashes.
When browsing a Reddit thread about a video game, if you encounter numerous comments written perfectly, especially those containing indicators like em dashes, the word delve, or similar language, it certainly can raise the question: am I genuinely seeing comments from users who write this way in this specific context, or is this content more likely produced by an LLM?
It depends. Em dashes in news articles and written publications? Definitely expected. Em dashes on social media or reddit? Either someone who works in typesetting, or an LLM. Most likely an LLM, giving the dying nature of printed media.
Only typography nerds and professional printers care about things like these. Popular media, even modern professional media, hasn't been paying all that much attention.
I’m not sure the same happened with “delve.” I saw an analysis of paper abstracts showing a clear uptick of “delve” starting with the mass-adoption of ChatGPT. Maybe it suddenly became a trendy word — especially in paper abstracts — or maybe more paper abstracts were edited by ChatGPT.
Combining the various "tells" of an LLM (em dashes, delve, grammatical signs etc) with the context (Reddit comments vs professional setting), you could establish a rough probability it was AI generated. At this point, it's the best we can hope for.
