Recently, there's been an influx of posts that have been entirely AI-generated in response to a question, particularly using the ChatGPT bot.
There have recently been some big advances in the capabilities of various AIs to generate content, including art and answers to questions. This is naturally exciting to us as AI enthusiasts. However, with these advances, there have been some negative side effects. One of those is writing AI-generated answers.

AI-generated answers, especially with the ChatGPT bot, can look very convincing at first glance. They're generally well-written, well-formatted, and sound like they know what they're talking about. They're convincing. However, upon closer inspection by a subject matter expert, it is often evident that the answer is blatantly incorrect and thus misleading. These "answers" are also extremely quick to generate; it takes a matter of seconds to copy-paste a question, recieve an answer from the bot, and copy-paste that into the answer field. (There have been thousands posted and deleted on Stack Overflow.)

The aim of Artifical Intelligence Stack Exchange is to provide a repository of high-quality questions and answers about the subject of artifical intelligence. Having misleading, mass-generated answers that sound high-quality but are actually incorrect runs counter to that goal. We are a site about Artificial Intelligence, not a site run by artificial intelligence; let's try to keep it that way.

  • 6
    $\begingroup$ How about just banning people posting misleading/incorrect answers, regardless of who/what wrote it? $\endgroup$ Dec 12, 2022 at 13:20
  • 8
    $\begingroup$ @FranckDernoncourt - If someone consistently writes low-quality posts, and show no signs of improvement, a suspension eventually will be issued. That is, however, a different problem than using a tool to automatically generate answers without even attempting to create quality content. One is simply poor contributions, the other is abuse of the site. $\endgroup$
    – Mithical Mod
    Dec 12, 2022 at 17:53
  • 1
    $\begingroup$ I don't think we can ever reliably prevent AI generated answers, but what we can do, is trying to ensure that the answers provided also cite or are made by a reliable source. It's of course very easy to just ban someone, but that is not improving the platform, and make it more reactive than pro-active. In this day and age with a majority of non-technical people asking and wanting to answer, this would become a problem, and a rather negative such. E.g. even some of the creators of MML's doesn't know exactly how it works, and it's behaviors. I think we need to be a bit more flexible here. $\endgroup$
    – not2qubit
    Mar 19 at 12:56


You must log in to answer this question.

Browse other questions tagged .