Wikipedia talk:AI generated content

Comments on proposed guideline

How is this different from what's required of any other articles? -- Auntof6 (talk) 09:49, 20 November 2024 (UTC)

It's not really other setting out specifically that AI content is allowed and saying what the copyright status would be. fr33kman 10:20, 20 November 2024 (UTC)

Many people right now are confused about how we deal with AI content. I think this page will help clarify that. However, there is much information lacking, and a proposal to modify the QD reason or to create one is required to effectively manage the content, which if we do not regulate now will be much more difficult to handle later. Some examples, some keys about identifying the AI content, who owns the content under different circumstances (like when the text or media closely resembles someone else's work) etc.. need to be added. BRP ^ever 10:33, 20 November 2024 (UTC)

I've added the QD reason but I need to contact the WMF legal counsel to have them explain the ownership issue. My thoughts are that WMF would end up with ownership. Thx fr33kman 11:12, 20 November 2024 (UTC)

OK, I've emailed the general counsel and will await his response. fr33kman 11:26, 20 November 2024 (UTC)

Different opinion

I always try to sound somewhat neutral, professional but chill on Wikipedia, but I'm sounding very opinionated this time, because I don't know how to word this better and I am bad at setting the right tone, so…

On WP:Simple Talk, Fr33kman said Do we want to allow this? […] I, for one, feel we should have a policy prohibiting it use. here and my answer is "No, I don't want to allow this". I am for banning LLM generated content, not for regulating it. I do not want to read an article that nobody wanted to write. If I wanted to read some LLM stuff, I'd ask chatgpt or something (it's never going to happen). AI fans don't look up things on Wikipedia anyway, because they have their LLMs stealing the content from Wikipedia and other sites and feeding it to them. Why are we allowing this? I don't want to allow this. This is not to say "I don't like it so I decided that we ban this" and I already know I am going to lose this "battle", but I'll still try to speak my mind. All I'm doing here is answering a very good question (if we want to allow this) and my answer is "no". If we allow LLMs, then why should I keep contributing to this project or any project ever? So demotivating. Uninspiring. The beauty of Wikipedia is going to be lost. Do people even care anymore? So sad. ✩ Dream Indigo ✩ 23:04, 21 November 2024 (UTC)

Important note: Fr33kman might have a totally different opinion than mine, and it's fine of course. Here's the full quote to avoid confusion: Do we want to allow this? Problems I see relate to copyright issues as well as problems with complexity and context. I, for one, feel we should have a policy prohibiting it use. Sorry for the confusion. ✩ Dream Indigo ✩ 23:08, 21 November 2024 (UTC)

I totally agree with you. I don't want to allow it either, but I feel the genie is out of the bottle and I feel the only thing I can do is influence how it gets used. It's happening on commons and enwiki so it'd be hard to ban it here. It's a sad day!! fr33kman 23:10, 21 November 2024 (UTC)

It totally is a sad day!! Possibly a sad future too :( And I am so glad you agree with me :) I too feel like it's too late to stop it completely, but I could never forgive myself if I never said anything at all. I should try at least, even if I know that it's leading nowhere, even if the road is dark and I can't see where I'm going, I'm still trying. Even if it's a very small attempt, it's better than zero. I don't wanna go off-Wiki too much so I'll stop here :') Signed: a very real, flawed and human ✩ Dream Indigo ✩ 23:57, 21 November 2024 (UTC)

Opinion

My opinion on Fr33kman's comment, Do we want to allow this?:

I am strongly supporting the idea of banning AI-generated content on this Wikipedia. Pretty soon, articles nobody wanted to take the time to write will be all over Wikipedia and I refuse to read those kinds of articles (which is also why I have not read Baby shower). If I wanted LLM content, I could type a prompt on Gemini or whatever, and then I probably would be busy creating "articles" that I didn't want to create myself! I'm not sure if anybody cares or not, but this is a serious problem regarding the originality of our articles. This is a disappointment. ⯎ Asteralee ⯎ 14:23, 26 November 2024 (UTC)

It's a huge disappointment Aster, but I'm afraid it's here. We could hole a discussion about banning it but enwiki and commons are already allowing it. I see it as a major threat to our project. I'm trying to have some control for us by writing the rules. (sigh) :( fr33kman 14:28, 26 November 2024 (UTC)

It really is a threat :( But thank you for agreeing with my opinion :) Even though it is kinda late to consider banning artificial intelligence (considering that enwiki and commons allow AI). Thanks for creating the Wikipedia guideline, you're a real golden editor :) ⯎ Asteralee ⯎ 15:26, 26 November 2024 (UTC)

Template

Hello. Overall, I agree with the content of this proposal. I have also created a template for marking AI-generated content (Template:AI-notice), which I would like to include here as a method for marking such content. Thank you. BZPN (talk) 14:44, 26 November 2024 (UTC)

Agreed. We need to let the reader know that the context is from an AI and not our usual excellent human based content. The template looks good. I'd go ahead and add it's use to the guideline. Thx fr33kman 14:56, 26 November 2024 (UTC)

This is an example of an AI-generated page: Androgynous. Depextual (talk) 20:31, 26 November 2024 (UTC)

Yeah, that's a great example of AI-generated article that should be deleted in QD mode. BZPN (talk) 20:50, 26 November 2024 (UTC)

Important problem

@Fr33kman, I forgot one more important problem with the use of AI that needs regulation - the use of AI in community discussions. For example, the use of AI during RfP or other discussions here. I believe that this is very unfair and the user should receive an appropriate penalty (e.g. exclusion from the discussion). BZPN (talk) 19:43, 26 November 2024 (UTC)

How do you see AI being used in community talks? fr33kman 20:43, 26 November 2024 (UTC)

Just like we saw at Talk:Baby shower. Statements clearly generated by AI and used by the user. BZPN (talk) 20:47, 26 November 2024 (UTC)

(I meant before deletion of this page today) BZPN (talk) 20:47, 26 November 2024 (UTC)

Ah yes. We'd need to make a rule and get consensus on it. I think WP:AI is one place it could be listed. Why not bring it up on ST and see what people say? After that we could make it part of the guideline. fr33kman 20:53, 26 November 2024 (UTC)

Of course, you're right. I will try to write such a proposal in a moment. BZPN (talk) 20:59, 26 November 2024 (UTC)

@Fr33kman, I wrote my proposal here: User:BZPN/AI addon proposal. Feel free to modify it :). BZPN (talk) 21:24, 26 November 2024 (UTC)

BZPN, is this widespread enough to the stage we need a guideline on it? If there is only one case of it happening, it likely does not require any guideline changes. --Ferien (talk) 21:25, 26 November 2024 (UTC)

@Ferien, It seems to me that one case is a sufficient reason to prevent further cases. Besides, I think that it is better that there is regulation and there are no doubts than that it does not exist. BZPN (talk) 21:28, 26 November 2024 (UTC)

BZPN, we have long followed a history of only adding guidelines/policies where necessary, mostly following en, in the few cases our own guidelines don't exist. Another guideline for a single case would surely be needless bureaucracy? Furthermore, your proposal does not even suggest implementing it to talk pages as that case was. It exclusively covers RfP requests, And a RfP clearly using AI-generated content would just be rejected immediately anyway. I am starting to get the feeling that a lot of this fear over AI may be overstated, because this just has never been an issue. --Ferien (talk) 21:34, 26 November 2024 (UTC)

It actually suggest: Article talk pages with only AI-generated comments will be deleted, as they do not help real information. I also added information about the use of AI on the articles talk pages. BZPN (talk) 21:42, 26 November 2024 (UTC)

Vandalism or nonsense comments on talk pages are also unhelpful, yet it is simply reset like any other disruptive talk page would be. Why does AI have to be an exception to the rule? --Ferien (talk) 21:48, 26 November 2024 (UTC)

┌────────────────────────────────────┘
I agree with Ferien. The use of AI in talk pages is much less of an issue than it is in the main space (articles). In the Baby Shower talk page, you could have reset the talk page and be done with it. Content in talk page can be removed if it doesn't pertain to improving the article's contents, and if it's a user simply repeating the contents of the article by having an AI explain in, then it's not really contributing towards improving the article and can be removed. No extra policies needed.

However, it cannot be used in community discussions with at least 3 participants. seems arbitrary. What happens if somebody posted something from chatGPT and it gets no replies for months? Would that make it fine since there's only one participant? — *Fehufangą ♮ ✉ Talk page 03:26, 27 November 2024 (UTC)

• For now, I'll just leave this proposal in my userspace. If you really want to wait, no problem. Maybe in some time this proposal will find reference in real cases of such use of AI. Thank you all for expressing your opinion. Have a nice day! BZPN (talk) 09:09, 27 November 2024 (UTC)

  Here we have another case of using AI in the discussion: User talk:106.221.147.224. BZPN (talk) 08:49, 30 November 2024 (UTC)

My opinion

I'm in favor of banning AI-generated content. Every AI-generated article I see goes completely against the purpose of this encyclopedia, to be simple. It's always either complex, written like an advertisement, or just completely false. Allowing AI-generated articles on this wiki is just asking to be flooded with unverified, complex content. Wikipedia should only include information that is verifiable. This wiki already has a big problem with spam and allowing AI would make it so much worse. FatalFit | ✉ | ✓  16:23, 30 November 2024 (UTC)

AI content is not ideal. We already have a half dozen AI articles up for RfD at the moment. We must come up with rules for it. fr33kman 16:47, 30 November 2024 (UTC)

Opinion on ai generated content

honestly i dont like articles that are generated by ai because:1 no sources 2 too much promotional phrases 3 sections that have bad organization and 4 no infoboxes i prefer creating articles manually using a english Wikipedia article as base but changing words to avoid copyright and also adding sources and of course adding neutral and non promotional words. 179.109.143.159 (talk) 00:38, 22 December 2024 (UTC)