It made sense in context
I've been playing around a bit with using large language models to proofread my blog posts. My conclusions so far are:
They are very good at spotting grammatical errors. I don't make a lot of spelling mistakes and I use a state of the art spellchecker to catch the few that make it through. But when reading back my own prose I just read what I meant to write, and the little bits of debris that are left over after rearranging a sentence while editing just seem to disappear. LLM's are great at catching that kind of thing.
They occasionally come up with a genuine insight into stylistic and thematic patterns that I wasn't consciously aware of. This is both satisfying and a little creepy.
They are absolute dogshit at editing or writing text. "I've rewritten your highly personal recollection in a style more suitable for a corporate press release. You're welcome!" No, thanks.
Whenever I succumb to the temptation of entering into a dialogue, things get surreal pretty fast. Here's a sentence that I got Google Gemini to produce:
Completely fair. There is a fine line between "enigmatic darkwave occultist" and "accidental spokesperson for 8th-century imperial dairy products".
I have to admit this is a fine line I was not previously aware of.