In my facebook community (the one I’m working to bring over to Discourse) we have started to see that. It’s extremely annoying, in a support community where relationships are fundamental to our work, to have people passing off AI slop (mostly) as their responses. I actually set up a dedicated chatbot with NotebookLM and all our documentation plugged in (proof of concept at this stage) for people who want to interact with an AI.
I get that not all communities have the same needs regarding authenticity and AI intrusion, but just to say that in mine, we are firmly in the “no AI, or if you want to quote something a chatbot told you, go ahead and quote with quotes and intro text, but really, you probably don’t want to do that”.
As @Canapin’s story shows, people get very annoyed when they realise the person posting has been dumping on them the mental load of sorting through whatever the AI spat out.
(For what it’s worth: I’m super excited by AI/LLMs and their possibilities, use them, but if I’m having a conversation with somebody I don’t want to realise that at some point in the discussion I switched to talking with their aunt, their dog, or their chatbot, without my consent.)
I am very favorable to this type of labelling. Transparency is important. AI is more than just a spellchecker or predictive text.
Adding: if this labelling is in place, it should be done in a way that posts can be styled differently to make the AI-ness very visible. E.g. different font, background, color, etc.