ChatGPT answers not permitted

Question

I am posting this as a new policy for ServerFault, but it has not yet been discussed among users or moderators of the site. It is a starting point for a policy, I fully expect it to evolve based on discussion here.

But as far as I'm concerned, answers generated by an AI such as ChatGPT are not appropriate here and will be dealt with expeditiously: answers will be deleted and accounts will be suspended.

Discussion on meta.SE about this.

Master list of policies on various SE sites.

Other recent questions about ChatGPT on meta.SE

I don't have - and I don't think the other moderators have - a lot of time to investigate possible AI posts, but at this point there are two signs that seem pretty reliable and will be used to decide if action is needed:

• multiple posts, fairly lengthy, often on disparate subjects, in a short period of time
• a positive score on one or both of this AI detector from OpenAI or this other detector from ZeroGPT.

Depending on how discussion on this post goes, moderators will be enforcing this policy when AI generated posts are detected - if you see posts that seem to fit, flag them and moderators will check, delete answers, and suspend accounts.

Answer 1

I welcome such discussion as in my view such answers fall into the plagiarist’s category. (As the bot feed itself upon open source answers like on here to complete, but it give no attribution)

For such category, plagiarist, I know myself and community member tend to detect them when new, low rep users post outstanding answer. Its a common red flag to check the answer background. I thank you for the AI detection link which is handy to use to detect the AI portion.

If the community want to deal it like plagiarist I welcome the discussion

Answer 2

As pointed out in a comment, questions written by ChatGPT also need to be considered.

So far, the only ChatGPT question I've seen is the one that's pointed out in the comment: serverfault.com/questions/1125632/how-does-one-fix-these

Although the OP was up-front about their use of ChatGPT, and their motivation for using it in this case made some sense, a large amount of time and effort was put in by multiple users in order to figure out what the actual question was actually about. I think there's an acceptable question in there somewhere, but even after a few edits it's tough to figure it out, which is highly problematic.

So just as the informal policy on ChatGPT answers is that they'll be dealt with quickly and harshly, ChatGPT questions will also be dealt with - maybe a bit less harshly if they seem to be reasonably motivated.