Timeline for Ban ChatGPT and AI-generated answers on Ask Ubuntu?
Current License: CC BY-SA 4.0
21 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jan 2, 2023 at 21:46 | comment | added | NotTheDr01ds | If we're going to "ban" ChatGPT here, it needs to be clear where the "line" is in terms of what is acceptable and what is not. | |
| Jan 2, 2023 at 21:46 | comment | added | NotTheDr01ds | @terdon As I mentioned in AUGR, you and I (and others) are in agreement on this, but it still does have to be stated explicitly, unfortunately, since there are those who do want to ban all use of ChatGPT, regardless of whether the usage is "assistance" or "generate/write." You are also correct that the use of AI is very difficult to detect in these cases unless disclosed. And we should encourage users to disclose the use of ChatGPT in these cases without fear of being banned (as long as the usage is "responsible"). | |
| Jan 2, 2023 at 18:20 | comment | added | terdon | I really think this is completely besides the point. Of course the kind of use of chatGPT (which by the way is not an AI but is just a language model stringing words together with no attempt to actually understand them) is fine. However, the kind of use you describe (using it to enhance already good answers or improve writing) wouldn't even be noticeable. We can safely ignore that since it isn't an issue and cannot really be detected. The point is about people who do not write their answers and instead present the bot's work as their own. | |
| Dec 31, 2022 at 15:10 | comment | added | MrPotatoHead | Great points @NotTheDr01ds. Agreed. | |
| Dec 28, 2022 at 14:23 | comment | added | NotTheDr01ds | @MrPotatoHead I do agree with that overall, but at this stage it might be safer to leave the wording as "should not be used to write ..." to prevent the misuse we are starting to see. For instance, one user posted 13 answers in the span of two hours yesterday; clearly not vetting them. However if, as we both agree should happen, a SME vets the output and explains that (and likely how) they did so, then I think they (can) ultimately transform it into their answer. | |
| Dec 27, 2022 at 15:32 | comment | added | MrPotatoHead | I don't agree that it can't be used to write answers. I would say regardless of whether it creates the answer or "improves" an existing answer, at this time its responses need to be vetted by a human SME. I say this from the perspective that I've given it code problems described in conversational English, and it's produced very good responses. This is from first-pass or dual-pass attempts, not where someone keeps providing instructions over and over to refine an answer. Point is, ChatGPT can be good at helping get through code writer's block, but I would not rely on it for the final answer. | |
| Dec 20, 2022 at 4:28 | comment | added | C.S.Cameron | Well said @NotTheDr01ds, but it is obvious that you do not need ChatGPT. | |
| Dec 19, 2022 at 23:32 | comment | added | andrew.46 Mod | @NotTheDr01ds We are indeed living in interesting times :). | |
| Dec 19, 2022 at 23:28 | comment | added | NotTheDr01ds | @andrew.46 As an example of a "bad actor", see this meta.superuser.com comment. I fully agree with banning that user, who was indiscriminately generating answers with ChatGPT. But I would like to see the tool used responsibly by users who are capable of doing so, without fear of being banned. | |
| Dec 19, 2022 at 23:22 | comment | added | NotTheDr01ds | @andrew.46 Right (and thanks for the kind words!) - But a "total ban" impacts both experienced users and inexperienced, whereas developing a policy and guidelines for usage allows for sensible usage, and still allows for banning the "flag fodder" for violating the policy. Win-win. | |
| Dec 19, 2022 at 23:17 | comment | added | andrew.46 Mod | @NotTheDr01ds Your use case sounds wonderful,however you are a dedicated AU user who posts well researched and carefully thought out answers. You respond to queries, you encourage new users and ideally you could be cloned! What percentage of users do you think would use ChatGPT in this way, and what percentage would instead unleash a tidal wave of ill considered, un-researched, crappy AI answers across AU in a frantic game of accumulating rep? I will go with 10% sensible use and 90% flag fodder... | |
| Dec 19, 2022 at 22:51 | comment | added | NotTheDr01ds | You and I seem to be in full agreement on the proper usage. It's just that we need any policy to be clear on it as well. | |
| Dec 19, 2022 at 22:50 | comment | added | NotTheDr01ds | @Levente It's a fine line, but the grammar example I gave was almost verbatim edited by ChatGPT. I would call that "ChatGPT-generated content", even if the input was an answer already. My point is that any ban should examine the corner cases and allow common-sense usage of ChatGPT. It doesn't seem to me that the current policies do so. I'm proposing a policy here that (I hope) attempts to cover the "do's and don'ts" of using ChatGPT on Ask Ubuntu (and other sites). | |
| Dec 19, 2022 at 22:45 | comment | added | Levente | O.k., I would not necessarily classify your example with the Windows-shortcut either as "posting ChatGPT content". In that example you are not crossposting output from ChatGPT to here, rather, you are augmenting your own knowledge with suggestions from the AI. But you vet and verify that knowledge, perhaps even rephrase it such that it is truly faithful to how you interpret it. The key takeaway is that — at least, for now — we want your ideas, not the machine's. | |
| Dec 19, 2022 at 22:36 | comment | added | NotTheDr01ds | @Levente I agree with the intent, but the suggestion from the original question is "Deletion of ChatGPT content, whether acknowledged as such or not". That reads to me like a "total ban", which goes too far, IMHO. That is the position of Stack Overflow at the moment. Any use of ChatGPT is prohibited, which is overboard IMHO. | |
| Dec 19, 2022 at 22:33 | comment | added | Levente | I think OP is not concerned about the language/readability enhancement-kind of AI usage. I think OP, just like Stackoverflow are concerned with cases where someone copies an entire question into ChatGPT, and posts the output here. That's the part where it can be wildly worthless (while looking absolutely convincing). At the same time, the part with the anecdote about Windows' window position and the new suggestion explored through ChatGPT is very interesting and thought-provoking. | |
| Dec 19, 2022 at 22:13 | history | edited | NotTheDr01ds | CC BY-SA 4.0 |
More thoughts
|
| Dec 19, 2022 at 21:09 | history | edited | NotTheDr01ds | CC BY-SA 4.0 |
Additional thoughts
|
| Dec 19, 2022 at 21:00 | history | edited | NotTheDr01ds | CC BY-SA 4.0 |
added 41 characters in body
|
| Dec 19, 2022 at 20:52 | history | edited | NotTheDr01ds | CC BY-SA 4.0 |
added 1595 characters in body
|
| Dec 19, 2022 at 20:27 | history | answered | NotTheDr01ds | CC BY-SA 4.0 |