What is the current view of the Unity forums moderators when it comes to the Chat GPT and AI generated answers in general? I think this should be addressed before the forums are totally polluted by these answers.
I wouldn’t want to see stuff like this below appear in quantity here, see one of the answers in the thread (it’s quite obvious:) How to enable Modding for your game
The LLM really can’t come up with anything new, it just reshapes the information it has scraped and in many cases adds totally made-up stuff too, which is not great. Stack Overflow already banned (at least temporarily) Chat GPT answers in early December last year (2022.)
Anyway I don’t know what the motivation of these users is who come here to post the generated answers. And maybe some of these fresh accounts are already some sort of Chat GPT API tests by someone, but it doesn’t really matter since I think the answers here should be by real users. And this will deteriorate the forums even more, since there are already enough issues as it is.
While I agree that these fake AI answers should probably be prohibited in some form, neither of those answers are chatbot generated. Both of those are active users who post sincerely.
those are automated bots, so should be considered as spam…
even if sometimes they do have correct context in the generated message…
(but completely miss the actual question, at least so far)
pretty sure they will eventually start adding SEO links to those replies…
It seems like a common strategy for spammers is to age accounts for a while by posting seemingly “normal” replies without spam links, in order to get around potential age/post restrictions.
Those used to be very transparent (“Thank you this helped me a lot”), but with AI-powered bots, you can chatbot generate messages for a few days, and then post your spam.
The funny thing here is that it’s completely useless on the Unity forums, because there’s no proper age or post restrictions - spam posts are generally from accounts with 1 post, with a ton of links.
The post was already deleted from the thread when you answered (it seems). It was a lengthy typical Chat GPT answer, with an introduction part, steps listed with numbers and a few reminders in the end… It was by a user with exactly 1 answers so far. And very obviously a Chat GPT answer. I guess I should have taken a screenshot.
@mgear I think my question was quite clear, maybe I should rephrase or try to express this more clearly. I don’t want to see machine generated junk answers. I want to see answers from users who have actually tested whatever-it-is themeselves. It boils down to trust I guess - I would like to imagine that there’s a real person somewhere on this planet who put effort in to figure out something and he can then tell someone else how to deal with the issue. Does that make any sense? This is a bit of a same kind of issue as dealing with some bank or webshop bots and their nonsensical answers. Try asking almost anything even semi-complicated or less common things and Chat GPT will happily fabricate you a convincing bunch of gibberish. I would rather not see that here, or at least have it clearly tagged to be some machine produced answer.
Not a mod, but as a language mo…, I mean forum participant I wouldn’t want AI-answers here as a norm. It is fine to direct a newbie toward a bot, but generally people that want to talk with a bot, should go to the site that hosts the bot.
Machine answers are confident, but occasionally hallucinate quite badly. I think that in technical fields specifically human and machine answers should be clearly separated. Meaning robots should not be allowed to answer in human section, and humans should not be allowed to answer in robot section.