Talk:Artificial intelligence in Wikimedia projects

Undue weight

I am publishing this draft with two sections - AI for Wiki and Wiki for AI. I gave more weight to the AI for Wiki section just because that is the concept which has the attention of the popular press and popular science discussion.

The weight of the sources is not in the popular press, but actually in the academic literature. There are many more sources talking about the use of Wikimedia projects for off-wiki AI projects than there are sources and projects applying AI to develop Wikimedia projects. I wanted to establish this article to be more accessible to the first wave of readers whom I expect to want to read this content and sort their thoughts about it.

Wiki for AI is much bigger now and the foreseeable future is that Wikipedia content will either be the basis of future AI research or otherwise the future of AI research will have a basis in Wikimedia content branded and further developed as some next generation dataset.

I would like to see this article re-written to identify whatever review articles exist and to give fair weight to the various topics which papers most deeply examine. Blue Rasberry (talk) 17:04, 24 August 2018 (UTC)

Wikipedia as a data set

• Mehdi, Mohamad; Okoli, Chitu; Mesgari, Mostafa; Nielsen, Finn Årup; Lanamäki, Arto (March 2017). "Excavating the mother lode of human-generated text: A systematic review of research that uses the wikipedia corpus". Information Processing & Management. 53 (2): 505–529. doi:10.1016/j.ipm.2016.07.003.

This article identified 132 other academic articles which describe how they used Wikipedia as a data set at the base of other research. As I start this article, there is a subsection on how artificial intelligence projects uses Wikimedia content in their development. That subsection could be split off into its own article, and that split article could itself be split into other articles. One such possible article could be something like "Wikipedia as a data set", because there is this article and the 132 it identifies as reliable source material for developing this as an independent concept. Blue Rasberry (talk) 16:58, 24 August 2018 (UTC)

"Artificial intelligence" as a buzzword

"Artificial intelligence" is a term with many meanings in various fields. Perhaps most of the sources which talk about artificial intelligence in Wikimedia are talking about machine learning, which often is another name for artificial intelligence. Blue Rasberry (talk) 17:12, 24 August 2018 (UTC)

I fully agree with you= AI is a buzzword. However, in the context of this discussion most of the time, it means generative large language model: https://cset.georgetown.edu/article/what-are-generative-ai-large-language-models-and-foundation-models/ .

The second most common meaning is generative large models for images : https://web.mit.edu/phillipi/www/writing/generative_models_of_images/index.html ApoieRacional (talk) 13:25, 29 October 2025 (UTC)

Bias study

There is no research reporting from this yet but here is a 2018 research project analyzing bias in Wikimedia data structuring.

• https://www.nsf.gov/awardsearch/showAward?AWD_ID=1815507

The PIs on this Brent Hecht and Loren Terveen seem to comment on Wikidata.

Blue Rasberry (talk) 21:28, 3 December 2018 (UTC)

ClueBot NG

Should a section about User:ClueBot NG be part of the article? The userpage states it detects vandalism with bayesian classifiers, a neural network, and a calculated threshold, but I'm not sure if that is "true" artificial intelligence. 172.112.210.32 (talk) 15:14, 22 December 2021 (UTC)

Yes, it definitely should be. PS also see section #"Artificial intelligence" as a buzzword. Prototyperspective (talk) 20:50, 13 December 2025 (UTC)

Just fyi the references from the ClueBotNG article failed verification with the text they supported (or were non-notable) ~2025-40581-55 (talk) 22:25, 13 December 2025 (UTC)

Is "Detox" real?

Does this actually do anything? I've literally never seen a user blocked, or an edit prevented/flagged, by anything called "Detox". It's currently described as "Detox is a project to prevent users from posting unkind comments in Wikimedia community discussions" -- this doesn't seem true to me. jp×g🗯️ 18:14, 2 February 2024 (UTC)

I added ‘update needed’ to it but if nothing is forthcoming, might be worth deleting ~2025-40086-48 (talk) 23:13, 11 December 2025 (UTC)

deleted per WP:AGEMATTERS ~2025-40305-16 (talk) 22:24, 12 December 2025 (UTC)

Merge

Do you think it was a good idea? Pxldnky77 (talk) 21:39, 25 August 2025 (UTC)

The subjects are different, one is about Wikimedia Foundation using AI tools to develop their projects and other is about users using AI to generate text on Wikipedia. I think they should be seperate articles. G13 vs G14 (talk) 12:05, 26 August 2025 (UTC)

The Wikimedia article was pretty sparse and shared some information with the AI on Wikipedia article (which had its own issues with bloat, as you can see I tried to remove some and added a warning. The article was a favorite for biased anti-AI Wikipedia editors. It might have been just as small without all the cruft.). Even though the Wikipedia Foundation using AI for their projects is notable enough to warrant an article I think it's better to just combine the two into one article on how AI relates to Wikimedia projects, as the individual topics aren't that different and don't have that much to say about them as of now. Pxldnky77 (talk) 18:29, 26 August 2025 (UTC)

Alright then G13 vs G14 (talk) 19:15, 26 August 2025 (UTC)

"exorbitant amounts of anti-AI info"

@Pxldnky77: I'm changing the {{Cleanup}} template to the more appropriate {{POV}} one which I think better reflects what you're trying to get across. Could you identify some examples of the "anti-AI info" in the article? I'll admit I missed the merge discussion and didn't know about the other article that was merged into this one, so don't know if there is other context that I am missing. ClaudineChionh (she/her · talk · email · global) 09:30, 27 August 2025 (UTC)

I have looked over the article and I do not detected a problem deserving of a POV top tag, and I do not understand what the alleged problem is. Per WP:WTRMT, I have removed the tag. —Alalch E. 10:00, 27 August 2025 (UTC)

I'm not really talking about the POV of the article. When I made the merge, there was a ton of information on the negative aspects of AI that seemed uncited or unnecessary. I cleaned up what I could but kept the tag to be safe. Pxldnky77 (talk) 19:57, 27 August 2025 (UTC)

Also I didn't discuss the merge because the two articles were so similar it didn't seem controversial. Pxldnky77 (talk) 19:58, 27 August 2025 (UTC)

Updating the ORES section?

I know that ORES is being deprecated and there's a move to LiftWing instead. Perhaps someone with a better understanding of all of this could update the section on ORES? -Kenirwin/(talk) 21:22, 3 November 2025 (UTC)

Agreed - added ‘update needed’ flag for now ~2025-40086-48 (talk) 23:14, 11 December 2025 (UTC)

The reason why I didn't add info about this last time I checked is because there was no reliable source about it available. This is why I only added the external link to LiftWing to the external links. Prototyperspective (talk) 20:49, 13 December 2025 (UTC)

I just removed reference to it altogether and described its purpose in helping to combat vandalism. Would be great to have more recent sources discussing it though (I also had a hard time finding them outside one-off mentions in academic articles) ~2025-40581-55 (talk) 20:58, 13 December 2025 (UTC)

"ORES" listed at Redirects for discussion

The redirect ORES has been listed at redirects for discussion to determine whether its use and function meets the redirect guidelines. Readers of this page are welcome to comment on this redirect at Wikipedia:Redirects for discussion/Log/2026 February 13 § ORES until a consensus is reached. Jay 💬 10:11, 24 February 2026 (UTC)

WP:LLMARTICLE

Update needed. This would be a case where some degree of self-reference becomes necessary. Ben Aveling 12:34, 24 March 2026 (UTC)

Edit request

−

In March 2026, English Wikipedia prohibited the use of AI to add content to articles, with exceptions for copyediting one's own writing and machine translation from another language's Wikipedia. The English Wikipedia community holds the position that LLMs often violate core content policies regarding verifiability, neutral point of view, and no original research.

+

In March 2026, English Wikipedia prohibited the use of AI to add content to articles, with exceptions for copyediting one's own writing and machine translation from another language's Wikipedia. The English Wikipedia community holds the position that LLMs often violate core content policies.

The source doesn't mention these specific policies (instead only saying "Text generated by large language models (LLMs) often violates several of Wikipedia's core content policies," Wikipedia's editing policy reads.), and, ironically, this list of three "core content policies" is very often found in... AI models writing about following Wikipedia's rules.

To clarify, I'm using the {{edit COI}} template here as I've taken part in several interviews regarding this guideline change, even if they're not directly cited by that particular source. Chaotic Enby (talk · contribs) 21:16, 27 March 2026 (UTC)

Done Delicious irony! Discourses on Livvy (talk · contribs) 23:17, 27 March 2026 (UTC)

source

• https://www.newyorker.com/culture/the-lede/the-accidental-architect-of-the-internets-brain

--Another Believer _(Talk) 18:27, 31 July 2026 (UTC)