Does the policy of banning ChatGPT answers on StackExchange sites exist here as well? If so, what to do when a user seems to be using ChatGPT?

I ask because a recently-joined user (3 days ago) has been answering questions, some of them from new users who joined three days ago (and thus may indicate gaming the system in other ways), in a way that clearly indicates GPT authorship.

These include this PTIJ question and answer, this question about the tree site icon, and rabbis arguing with God. This other answer is to a new user but doesn't shout out as being particularly AI generated.

Some sample text that indicate LLM authorship, e.g. in the tree question:

Overall, the passage intricately explores the dynamics of human consciousness, the intertwining of good and evil, and the process of spiritual purification necessary for eternal life. It weaves together philosophical reflections with mystical insights, offering a rich tapestry of understanding.

2 Answers 2


I don't see any reason why Mi Yodeya should be an exception to the general StackExchange policy against AI-generated posts.

You seem to know more about LLMs than I do, Josh, but I know enough to confidently say that the positions outlined in the policy post that you cited are reasonable concerns about LLM content, and sufficient reason to ban them on MY.

Overall, because the average rate of getting correct answers from ChatGPT and other generative AI technologies is too low, the posting of answers created by ChatGPT and other generative AI technologies is substantially harmful to the site and to users who are asking questions and looking for correct answers. [...]

As such, we need to reduce the volume of these posts and we need to be able to deal with the ones which are posted quickly, which means dealing with users, rather than individual posts.

So, for now, the use of ChatGPT or other generative AI technologies to create posts or other content here on Stack Overflow is not permitted. If a user is believed to have used ChatGPT or other generative AI technologies after the posting of this temporary policy, sanctions will be imposed to prevent them from continuing to post such content, even if the posts would otherwise be acceptable.

For what it's worth, I also agree with your assessment that many of the posts that you cited as examples of possible AI-generated posts are likely to be AI-generated, in addition to being general low-quality posts.

  • But what about all of the low quality answers that people write, should they be banned also? Commented Mar 26 at 7:57
  • 1
    @Awtsmoos--עצמות If they are that bad, then yes.
    – Double AA Mod
    Commented Apr 2 at 16:35
  • @DoubleAA then why is the rule simply not "no bad answers"? Ai can write really good answers. It delends how it's used Commented Apr 2 at 18:50


In your quest to sniff out AI like a bloodhound, maybe take a moment to consider that humans can write intricate passages too. Just because something sounds smart doesn't mean it's automatically AI-generated. And FYI, AI is just based on human data, so if it sounds human, it's probably because humans wrote the same thing. Now, why don't you mind your own business and let people answer questions however they want?

  • 4
    It isn't that it is intricate or smart. There is a signature writing style and structure. As to minding my own business, there are drawbacks to allowing AI generated content, such as hallucinations quality, and that people come for human answers. They can use free GPT themselves to look things up, but a human might actually think things through. Commented Mar 20 at 22:49
  • @joshwaxman the fact that you didn't suspect this answer of being written by ai proves everything :) Commented Mar 21 at 0:59
  • 1
    If you did, it also proves nothing. A short snippet can be generated in all sorts of styles, and doesn't always have the telltale signs. But your other answers do, in fairly known ways. As I explained to my students in the LLM class I just taught, there is a difference between precision and recall. recall is, of everything in the class, what percent do you find? Precision is, of those you classify as a certain class, what percent do you get right? Commented Mar 21 at 1:26
  • @joshwaxman if you want to down vote something because you disagree with the quality that's your choice. But ai could literally be anything. It's just based on human input, the question is how good is the input Commented Mar 21 at 1:50
  • 6
    I'm not sure I understand what you are trying to say, or that you understand how LLMs work. But regardless, they are banned on Stackexchange sites, so you should not join such a site as a new user and use them to make your answers. Commented Mar 21 at 1:53
  • @joshwaxman I myself identify as an llm. The entire universe is ai, deep down. Rules that have no meaning are automatically void. Why don't they just have a rule to not have "low quality" answers, if that's the whole problem? Because there's no way to define it. So too, all people are ai deep down, so there's no way to limit it to anything Commented Mar 21 at 2:07

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .