There was a blog post that reached first page here on HN a few months ago. IIRC, only one commentator alluded that "the post is so nonsensical as if it was written by GPT-3, no solid arguments just mishmash of phrases"...surprise, it was written by GPT-3.
(BTW, I'm not suggesting anyone here is a bot, obviously)
I used GPT-3 to generate responses to someone who argued a post is GPT-3 generated on an old account. It was pretty funny - it fooled them, but smaller text is usually harder to detect as bot-generated.
(BTW, I'm not suggesting anyone here is a bot, obviously)