Lately, there seems to be a noticeable number of AI-generated answers in this forum. I couldn’t find this topic being mentioned in the forum guidelines, but I know that other sites have created policies for this sort of content (stackoverflow comes to mind).
My goal here is to help figure out what people here think about such content, discuss the pros and cons of it, and maybe have the feedback in this thread help shape policies or guidelines regarding this topic.
I’ll start by sharing my personal opinion:
LLMs can be a great resource. However, if I have a question where I think an LLM could be of good help (e.g. how do i write a custom SNMP check for a rare device to check X and Y? I have a MIB file.), I will ask my AI of choice directly.
When I ask on the forum however, I do it with the intention of getting answers by humans who likely gained their knowledge from experience. Because it has historically been like this, forum answers - to me - have a high level of quality and hence implicit trustworthiness.
AI-generated (even if only partially) answers can still be helpful (e.g. instead of “yes that is possible, see <link to docs> on how to create your own SNMP check”, you’d get a well-structured and well-formatted answer with example code), but it can sometimes also be harmful, namely when the facts presented are wrong, which as we know does happen.
When I ask an LLM, I know this, and I treat its responses with the necessary caution and skepticism, and I double-check everything.
When I get a response on the forum however, I expect the responses to be factual and not just guessed, and if it could be wrong, I expect the poster to mention this (e.g. “this worked for me on 2.2, but you’re on 2.4, so beware that things might have changed” - something where AI often uses outdated knowledge and when you tell it “I tried it but it doesn’t work, did you by any chance give me advice for the old version?”, it will say “oh, of course, you are completely right! sorry about that.”).
Personally, I don’t want to get to a point where I have to question the correctness of forum answers in the same way I need to do for AI answers.
Consequently, I think that AI-generated answers should either be banned or at least be required to be marked as such. There have been a few posts I’ve read that made me go “
” only to later realize that this must have been written by AI. So to avoid confusion, a disclaimer like “this answer contains AI-generated content” could be helpful. I think this would also be better than banning AI content completely because, well, it can also be quite good.
What are your opinions on this? Other ideas on how to deal with this?