Accepting/Allowing untested ChatGPT "answers"

Ideas & Suggestions
,
1

(Couldn’t find a related post)

What’s your or Athom’s opinion about people trying to help, by just posting untested ChatGPT (or similar services) “answers”?

I think it’s just annoying and useless.

7 Likes
2

I aggree,
Useing AI to generate an comprehensive answer could be faster and helpful but should be checked cq verified to be correct.

I also see many “help” from many Asian people with often comprehensive answers in their first post and with les than 2 minutes reading on the community after registering.

Imo the rules for using AI should be as with many companies:

  1. use AI responsible
  2. check the answers and report your conditions you checked it with. Fe: Firmware version and Mobile App version
  3. add a Note / source if you used AI and that you checked the answer

Note that probably 90% of that army of ppl that try to “help” are getting a Ban here in the first 12h. The rules and AI from Discourse is helping identifying these and ar often reported. An IP geolocation and some effort from the moderators here remove a hand full potential spammers every day.

Ps: sometimes I come over a old post with after re-reading an AI answer and edited day’s later with a external link. Pls is you encounter that flag :triangular_flag: that posts. We can check their IP and other accounts using that IP and otherr posts to cleanup and prevent other misuse of links.

2 Likes
3

That’s very typical, also on other platforms (I see it on StackOverflow a lot as well): pretend to post an answer (generated by LLM), then wait a bit, then change it to some sort of spam message.

2 Likes
4

this is also par for the course on Reddit too