|I totally understand their position and motivation, but my honest feeling is it's almost impossible to police properly, and is akin to them saying "don't cut and paste another person's answer". AI is very hard to spot, and the use of AI to generate a solution is no more morally suspect than someone using Google to spot a random blogger's note on the topic and use that as an answer.
Stackoverflow's reasons are: the answer may be noisy and over-explanatory, it may not be factually correct, it won't have correct citations, and if someone wanted an answer generated elsewhere (eg AI) they would have already searched. The first 2 reasons are already covered under the general "don't post bad answers", I rarely see citations on SO, and no, users don't look elsewhere first because StackOverflow has awesome SEO so they come up first in many searches.
If someone knowledgeable in the space uses AI to generate an answer on a topic they understand, checks that answer, trims it down to what's needed, and posts, then I'm 100% OK with that. Saying otherwise would be akin to saying we should ban code snippets generated by CoPilot. For me, the only thing that matters is
AI is here. We can't stick our heads in the sand about it, and we can't say "Just Say No". There's an onus on us to understand the tech, understand the implications, and understand and use it in a way that's positive while also facing the negative implications of AI. I don't feel a blanket ban does any of that.
- The answer is correct
- The answer is on topic and specific to the question
- The answer isn't plagiarized. (This is such a debatable concept when it comes to AI), and isn't simply a cut and paste from any source
- The answer isn't offensive, abusive, so poorly written it's not understandable