Just came across an interesting use case for large language models: assist with writing text in (human) languages that the user is not confident with.

I have in front of me a comment in flawless English, written by someone who I know can't write like that.

On the other hand, the text does read like it was written by an LLM, so this could hinder more than help, depending on the context.

(Also, the comment is not really very useful, but I suspect this is more about the prompter than the LLM).