Chatbots (and the large language models that power them) have become very popular; they can often output material much faster than humans can. Unfortunately, they cannot write as well as humans (yet), and their output is prone to hallucinations, false citations, and other errors. This has created a major cleanup burden at Wikipedia, as many editors (especially new ones) try their hand at using artificial intelligence to edit Wikipedia. You can help by identifying AI-written text, removing unsourced or inaccurate claims, and by identifying AI-generated images. For more information, see our AI Cleanup Guide.
LLM-generated essay, basically what you'd get if you asked a college junior to submit a summary of the topic citing no fewer than 10 references. Edit history and talk page reveal it once had an AI-generated graphic that was removed for inaccuracies; article creator replies on talk page also appear to be entirely LLM-written. I have to give the author credit for at least engaging with other editors on the talk page (rather than dropping this mess on us and abandoning it, like most LLM-wielding editors). But we can't have articles created in this manner per WP:NEWLLM. Delete per WP:TNT as there is a topic in here, but the article is unsalvageable. WeirdNAnnoyed (talk) 11:15, 2 June 2026 (UTC)[reply]
I appreciate the acknowledgement that I have engaged with other editors on the talk page - that was my intent. Furthermore, I have no intention of creating a quick-drop in article that will be abandoned. I have worked on this topic for more than a year, including revising the title, editing and removing material, correcting references, responding on the talk page, and making changes and requesting feedback.
The article was a also developed through a Wikipedia education course. During this course, it was reviewed and suggestions were provided for improvement. This is relevant because the article was developed through an extended learning and revision process.
I also want to respond to the characterization of the article as something like a casual college assignment. That description does not reflect how the article was developed. I am a published author in statistics and experimental development, I approached the article as a careful review of the published literature. I understand the value of reliable sources and would rather focus the discussion of policy-based concerns that can be addressed through editing.
Regarding the images. I accept that AI generating graphics are now problematic - I simply did not know before hand, hence the reason I was fully transparent in describing how they were generated. My intent was to create a conceptual illustration because I was unsure that I could copy images from a published paper - something I need to follow-up on. Again, I created these images and reviewed them during my course because images was a topic of discussion and I felt the need to create images to explain an important concept. But, as was explained to me by one of the editors on the talk page, they should have been generated from data. This policy was unaware to me. Furthermore, I still don't know if data generated knowing the parameters of a distribution would conform to wiki policy? I trust some here can tell me.
The nomination itself acknowledge that there is a topic here. If the editors believe the article is too board, insufficiently tied to the cited sources etc. I am willing to help changes and source-check. A Cleanup would be more appropriate than deletion. As such, I would ask that the issues of the article focus on specific issues with sourcing, neutrality, original research, article structure, rather than personal comparisons about the creator. Amlvjvch (talk) 13:43, 2 June 2026 (UTC)[reply]
I'm happy to see an acknowledgement that the subject indeed exists and thank you for pointing to one reference that suggests so. I would appreciate specific-based concerns tied to specific sections, in this article, that can be reviewed and fixed?
The previous AI assisted images raised concerns. Those images were removed and I did not object. In fact, it is was a learning process that I appreciate because it had a specific targeted outcome. However, it is discouraging, given the time revising this article, to see a blanket LLM accusation applied to the entire article. If there are specific targeted concerns with sourcing, article structure then please identify them and I would be happy to source-check and update content. ~2026-32794-26 (talk) 15:00, 2 June 2026 (UTC)[reply]
This should ideally go back through the Articles For Creation process (AfC), where it can be reviewed. While I'm not certain of the process, other editors would be happy to assist you. Oaktree b (talk) 19:51, 2 June 2026 (UTC)[reply]
Both of these sources even remain in the current revision of the article.
Going through the reflist of the current state of the article (at revision 1345449859):
Refs nr. 1, nr. 17, nr. 18, nr. 19, nr. 23, and nr. 24, return 404 not found.
Refs nr. 2, nr. 6, and nr. 8 is the BBC source that was present on creation and returns 404 not found (www.bbc.com/news/world-africa), with a couple of different titles.
Zambian radio engineer (BLP). Only two of the sources actually appear to exist, and these are both on the website of the subject's employer (the African Telecommunications Union). The other sources all return 404 errors, and on spot-checking these sources in the Wayback Machine I find that they already have 404 errors in the month that they were allegedly retrieved, which may suggest that these webpages never existed in the first place (?). I also see no evidence here that the subject meets WP:BASIC.
This article was almost certainly AI-generated, the same user that created it wrapped the wikisource in a Markdown codeblock in revision 1356294027, which chatbots often do when they are asked to produce wikitext. The leading ```wiki was removed in revision 1356294055, but the trailing ``` remains on the bottom of the page.
Draftify - Agree article is written in breach of WP:EPAI and needs to be rewritten and re referenced. I am however satisfied that subject meets notability in line with WP:NSUBPOL. BleachPuffin (talk) 16:19, 31 May 2026 (UTC)[reply]
Sure, but this has nothing to do with notability. I opened this AfD purely because the current article itself is a WP:NOLLM violation, not because of the subject of the article. Since it needs a TNT anyway, there is no practical benefit from draftifying and leaving the AI-generated text in the page history. That being said, I won't oppose anything that removes it from article space. --gurkubondinn18:05, 31 May 2026 (UTC)[reply]
Delete Unlike most weak human-created articles, LLM articles are an active detriment to future efforts because they sound so self-assured while being riddled with issues. Agree with Jumpytoo on this. M kuhner (talk) 03:25, 1 June 2026 (UTC)[reply]
Delete per excellent assessment of Nom. Recreation would need to exclude LLM violations. Note: The 33 "ref nuke bombings" on a single-sentence paragraph stretches the word "excessive". There are approximately 36 subnational entities comprising around 4,120 elected members in India, so advancing notability is important. -- Otr500 (talk) 03:50, 1 June 2026 (UTC)[reply]
Beneath the suffocating blanket of LLM bullet-point hell, this article is just a WP:DICTDEF. Of the three sources cited, one is unrelated to the topic, one is a dead link, and one is a blog post. A quick Google Scholar search uncovers several articles on naming conventions for particular items (e.g., drugs and biologics), but nothing on "products" in general, suggesting the topic is not notable. WeirdNAnnoyed (talk) 20:28, 30 May 2026 (UTC)[reply]
I want to agree, but I stand by my original call for deletion because product naming is a very weak, very vague single-source article. I also think "product naming convention" is a rather unlikely search term, but I don't have any evidence for that. If people are really set on a redirect, a better target would be Brand#Global brand variables which has much more content. WeirdNAnnoyed (talk) 12:39, 31 May 2026 (UTC)[reply]
Biting on this: Brand naming and product naming aren't actually one thing. They can be, but a corn chip is a product name, not a brand name, but some marketing person had to decide that this thing would be called a "corn chip." Or "shredded wheat." Or a "roller coaster." Hyperbolick (talk) 19:21, 1 June 2026 (UTC)[reply]
Original author here, I understand the LLM hits as I have had to pull my own websites down. For the purposes of this article it was meant (and maybe failed at) discussing the length, and predictable nature of a product name. I have no concerns with deleting. It has not gathered any real attention. Maybe a better person could summarize the whole article as one sentence on a parent topic. Lathama (talk) 21:16, 1 June 2026 (UTC)[reply]
Comment: I acknowledge the concern regarding the current reliance on primary sources. However, the organization is a legally registered Dutch charitable foundation that has been operating since 2012 and holds ANBI public-benefit status. I am currently reviewing additional independent sources and third-party coverage that may further improve the article.
At this stage, I believe the article would benefit more from improvement and additional sourcing than deletion. Concerns regarding sourcing, tone, or structure appear addressable through normal editorial processes. Phoeniks06 (talk) 21:22, 30 May 2026 (UTC)[reply]
Delete I concur that this is likely LLM generated and a violation of WP:NEWLLM. The article contains text/topics that no human would put in such an article, and only either an LLM or someone paid to make a marketing article would do. Regardless in either case the article needs to go. JumpytooTalk18:36, 31 May 2026 (UTC)[reply]
Comment: Thank you for the feedback. I understand the concerns regarding the article's initial reliance on primary sources and its relatively generic structure.
Since the nomination, however, independent secondary sources have been added, including coverage from NOS and Open Rotterdam concerning activities involving the organization. As a result, the article is no longer based solely on the organization's own website and social media publications.
Additional independent sources are currently being reviewed and will be incorporated where appropriate and in accordance with Wikipedia's sourcing guidelines. Any remaining concerns regarding tone, structure, or sourcing appear to be issues that can be addressed through normal editorial improvement.
Given that the organization has been active since 2012, holds ANBI public-benefit status in the Netherlands, and now has independent coverage included in the article, I believe that further improvement and sourcing would be more appropriate at this stage than deletion. Phoeniks06 (talk) 18:37, 31 May 2026 (UTC)[reply]
Delete By my count, 5 of the 8 sources are either dead links or do not go to the described article. A rate that high on an article just three days old strongly suggests LLM sourcing. There is also no attempt to show notability. M kuhner (talk) 06:56, 27 May 2026 (UTC)[reply]
Odd - when I declined the speedy I checked the references, and I just checked them again now. None of them are dead links, and although I can't read the non-English ones, the two that are in English are about the article topic, and the non-English ones that use Latin characters appear to be about the topic.
Also - I would argue that as notability is about sourcing, citing any sources at all, including unreliable or made up ones, is an "attempt to show notability". (My comment should not be taken as a keep argument) ~ ONUnicorn(Talk|Contribs)problem solving12:12, 27 May 2026 (UTC)[reply]
User:ONUnicorn on what planet is using made up sources acceptable to show notability?! Based on that criteria literally anything could made notable. I could write an article about my cat and just invent the fact that they were covered in dozens of news articles. Zackmann (Talk to me/What I been doing) 15:07, 27 May 2026 (UTC)[reply]
Delete - Even if a compliant article could be written on this subject, this draft is so LLM-infested that this continuing to exist would actively hinder someone from writing that theoretically better article. CoffeeCrumbs (talk) 19:15, 29 May 2026 (UTC)[reply]
AI-generated from first revision. e.g. the statement "The Italian Geological Society now highlight Villaggio del Pescatore as the key dinosaur‑bearing locality of the Trieste Karst" is not supported by the source. There's a lot of LLM-esque promotional wording such as "underscores" and "widely recognized". So basically delete per WP:NOLLM. --not-cheesewhisk3rs ≽^•⩊•^≼ ∫ (pester) 08:15, 29 May 2026 (UTC)[reply]
Redirect to Lipica Formation, the locality is documented as a paragraph there, as the locality is an outcrop of said formation what can be said about the locality is better but in the article in the formation. Lavalizard101 (talk) 11:29, 29 May 2026 (UTC)[reply]
Keep There is no significant evidence the article has been LLM generated. Please remember that mere stylistic signs are not reliable evidence of LLM text. I am Italian and if anything the text seems instead to be written by a native Italian speaker; see expressions such as "uncertain collocation", "contributing to debates" or "These represent the first late Cretaceous records" that are slightly awkward Italian calques, that would be unexpected in an English text produced by a LLM. That said, site is a massively notable lagerstätte discussed in several academic sources already present in the article, see [2], [3], plus other Italian language sources such as [4], [5], [6]. If someone needs an expert revision of the article, I know first hand some of the researchers involved. --cyclopiaspeak!13:21, 30 May 2026 (UTC)[reply]
If you had read my nomination statement thoroughly you would have seen that I did not rely on "stylistic factors" only. But if you need more convincing, here's other indicators of AI use, including what I already wrote above about the "key dinosaur-bearing locality", because you've ignored it.
Recent authors emphasize the need for continued protection and promotion of Villaggio del Pescatore as a unique geosite of international value[3][4]. Neither refs 3 nor 4 support this claim.
Villaggio del Pescatore's best‑known fossil is the hadrosauroid Tethyshadros insularis, represented by exceptionally complete, articulated skeletons.[1][5] Ref 1 does not support "exceptionally complete, articulated skeletons"
The Italian Geological Society now highlight Villaggio del Pescatore as the key dinosaur‑bearing locality of the Trieste Karst[4]. Not supported by source.
These 3 examples alone are evidence enough of AI use, as they contain evident LLM verification failures and synthesis of published sources. That's just a sample; there is more problematic content in the article than I have time to list. --not-cheesewhisk3rs ≽^•⩊•^≼ ∫ (pester) 14:17, 30 May 2026 (UTC)[reply]
Do you really believe that people didn't put references that happened to not support a claim before LLM? I'm here on WP since 2005, I guarantee that has always been pretty common. cyclopiaspeak!08:44, 1 June 2026 (UTC)[reply]
Viceversa, I find it utterly implausible that a LLM would output text with awkward calques typical of Italian speakers of English. cyclopiaspeak!08:45, 1 June 2026 (UTC)[reply]
Also, I have doubts you really checked the sources for the claims.
1. Ref 3 says right at the start "The Villaggio del Pescatore Lagerstätte (VdP hereafter), located in the municipality of Duino-Aurisina, Trieste (NE Italy), is a world-renown paleontological site due to its well-preserved Cretaceous vertebrate remains, standing out in the European palaeontological landscape" - so it supports the claim partially. Ref 4 doesn't, but it stresses the importance of the VdP fossils, and as such an inexperienced editor could have easily added it to strengthen the claim (see point 3 below)
2. Ref 1 describes multiple articulated T. insularis fossils, while Ref 5 describes it as "One of the most complete dinosaur fossil ever found". Yes, there is a bit of puffery/synthesis, but nothing unusual from an unexperienced editor.
3. I agree that the sentence is not fully supported by Ref 4. However, the reference states "The fossils collection is mainly representative of the Karst and Istria areas, and it is subdivided into historical collections and more recently acquired specimens. The most important collection is certainly represented by the Cretaceous fossils from the Villaggio del Pescatore site" - As such, again, it is a reasonable deduction from an inexperienced editor. It is also, incidentally, true: it is likely that the editor looked for a source that could somehow support what they knew to be a correct claim. Heck, I've been guilty of such clumsy editing when I started editing WP twenty-something years ago. That seems much more the output of an overenthusiastic but rookie WP editor than LLM hallucination. cyclopiaspeak!09:01, 1 June 2026 (UTC)[reply]
The article appears to be completely written by an LLM, due to it being in an AI's writing style. Many information is unsourced, and some is unconfirmed, like the paragraph stating that the original name of Wonderoos is "What-To-Doodles". Although it is sourced, and the source indeed shows that name, and it has a description similar to Wonderoos, there has never been any official confirmation from Netflix or 7ate9 stating that it indeed was the previous name of Wonderoos. RotatingPirateShip (talk) 15:50, 28 May 2026 (UTC)[reply]
Delete. The amount of unsourced material here is staggering and I see no reason to believe any of it is reliable given its probable origin and the mismatches with sources I checked. If someone wants to write an article on this show (there are enough reviews that it could be reasonable) this material will harm rather than help their efforts. M kuhner (talk) 17:31, 28 May 2026 (UTC)[reply]
I don't see it as reaching G15, because we don't seem to have Implausible non-existent citations or nonsensical references as G15 requires. They do, alas, exist. Absolutely clearcut AfD though. M kuhner (talk) 20:46, 28 May 2026 (UTC)[reply]
Keep No evidence whatsoever that article has been written by LLM (it seems to me a very normal stub); there are external journalistic sources so it seems to meet WP:GNG somehow. --cyclopiaspeak!12:51, 28 May 2026 (UTC)[reply]
Keep.
Which elements failed verification in the nomination diff? User:Euphomedia has been making edits to clean up some of the issues, and I corrected a typo. The newspapers seem reliable, as they have editors, but I can't verify the podcast. SenshiSun (talk) 13:24, 28 May 2026 (UTC)[reply]
Thank podcast is an external or extension link of the Radio interview for a book review that the author had with a prominent radio station in South Africa Ikwekwezi Fm. It is part of the SABC News. Euphomedia (talk) 13:49, 28 May 2026 (UTC)[reply]
Okay, thank you. I couldn't see that right away. All three sources look valid now.
Comment Are we seriously believing that v1 is not LLM generated? NBOOK requires substantial reviews. The reviews are not substantial. Fermiboson (talk) 21:07, 28 May 2026 (UTC)[reply]
That version is quite obviously LLM generated, as it is full of LLM formatting and weird text. LLM articles have been saved in the past with significant rewrites. When I checked the version available to me at the time, the only thing that failed verification was a typo. I will double-check the reviews for how substantial they are.
Personally, I would not be interested in taking advice about making the most of my 20s from someone who's only 23, but that doesn't factor into my editorial decision. SenshiSun (talk) 22:36, 28 May 2026 (UTC)[reply]
The first two articles are detailed, as they discuss the book, its author, and its target market. Neither offer any criticism or questions. I can't check the third one right now because it's a podcast. I will try to listen to it later. I will see if this has any public press releases with similar content to the articles. SenshiSun (talk) 22:50, 28 May 2026 (UTC)[reply]
Issues present on v1 (and there were many) aren't what this discussion is for, the important thing here is the current revision. All LLM issues seem to have been suitably solved by the re-writes done by others and everything passes verification Lovelyfurball (talk) 14:03, 29 May 2026 (UTC)[reply]
Keep, the reviews which are on the article show that it (albeit narrowly) meets WP:NBOOK. The reviews go into enough depth. The article seems to have started out as LLM-generated, but the current revision is perfectly fine. It is a bare-bones stub with every detail cited to a source. It has been re-written enough that deletion as a WP:LLM violation would be counter-productive. The ways we deal with LLM-generated content is either removal or a re-write, and the re-write has done enough to address any issues LLMs may have caused. Issues present on v1 aren't relevant since this discussion is about whether to keep the current revision. Lovelyfurball (talk) 14:01, 29 May 2026 (UTC)[reply]
Case has been raised against author at WP:AI noticeboard#Bøttle-x for LLM use. Spot-checking this list, of which they are original author and sole editor, shows multiple problem citations of typical LLM type including links that go to a different article than stated and sources that do not support the list entry they are attached to. Checking the entire thing would be an enormous amount of volunteer work for little apparent gain. There are 44 entries of which less than half go to extant articles. No attempt has been made to demonstrate notability. I am also concerned that there may be too-close paraphrase of entries at Aesthetics Wiki (not cited), from which many of the summaries seem to have been derived. M kuhner (talk) 03:34, 28 May 2026 (UTC)[reply]
As far as I can tell they did not admit to LLM for Internet aesthetic. They admitted to LLM for the current article and for Weirdcore aesthetic.I do not see the point of the proposed redirect in any case. Who will be looking for "list of internet aesthetics" and not "internet aesthetic"? M kuhner (talk) 00:50, 29 May 2026 (UTC)[reply]
Because "List of Xs" is an exceedingly common Wikipedia page naming scheme, and the target article does in fact have a list of aesthetics under Related examples? ~ A412talk!01:56, 29 May 2026 (UTC)[reply]
Weak Keep / TNT: The topic is notable enough so a list of internet aesthetics should probably exist on Wikipedia (just like List of subcultures), but the current article is LLM-generated which violates WP:NOLLM. Furthermore, a few of the listed aesthetics here appear to be hallucinated, like "Chaoscore." Miiversal (talk) 15:40, 1 June 2026 (UTC)[reply]
Article author has a case against them at WP:AI noticeboard#Bøttle-x for LLM use. It is difficult to check source/text matching here as key sources are paywalled (and not hotlinked) but in any case I struggle to see why we need an article on a single hieroglyph, notability is not shown or even attempted, and most sources are primary (lists of hieroglyphs). Writing style is very LLM-like and feels padded. I see no point in making a redirect. M kuhner (talk) 03:54, 28 May 2026 (UTC)[reply]
Additional comment: the phonetic value given the character looks like what you'd get if you eyeballed the symbols in the source and wrote down the closest letters you could find. I don't think it's correct. M kuhner (talk) 03:55, 28 May 2026 (UTC)[reply]
Comment - just on the sources, the Gardiner grammar is available to borrow via the Internet Archive, and I have added the URL and page number to the citation, and "Iconography of the Sledge in Ancient Egyptian Funerary Art" is pretty readily available on academia.edu via a search engine. It contains a decent paragraph of text comparing this character and another one. Cheers, SunloungerFrog (talk) 09:31, 28 May 2026 (UTC)[reply]
Userfy per editor's request (with a later deletion), provided that they won't it back into article space. Per their own comment at WP:AINB § User:Bøttle-x, this article was eyeballed and some info was written without a souce, so it should not be in article space. Hoaxes are generally not permitted in user space either, but I think it is allright to give them a bit of time to save a local copy or something, as a show of good faith for having been straight forward and honest with us about their LLM use (and said "eyeballing"). --gurkubondinn10:55, 28 May 2026 (UTC)[reply]
Delete as failing NOLLM. The user has gone right back to submitting more AI-generated material and thus there is no point in extending olive branches. Besides, they've had enough time to save a local copy by now regardless. --gurkubondinn16:07, 30 May 2026 (UTC)[reply]
Declined PROD was replaced by materially better article but there is still not enough to establish notability. AR/12News only cite the school district in the context of other Maricopa Co districts and even then, only as WP:ROUTINE coverage. Other sources do not assert notability. In contest of prod, author cites nonexistent WP:NGOVT policy, which makes me wonder if AI was used. Allan Nonymous (talk) 12:51, 27 May 2026 (UTC)[reply]
The following discussion has been closed. Please do not modify it.
Keep. The article meets WP:NORG through independent, subject-specific coverage. The nomination appears to discount the sources too broadly as routine coverage.
Policy frame: For a school district, the relevant standard is WP:NORG/WP:CORPDEPTH, not the notability guidance for individual schools. The question is whether reliable independent sources address the district directly and in some depth.
Sources analysis:
Arizona Auditor General, Arlington Elementary School District Performance Audit (2016) — an independent state audit report devoted entirely to this district. It discusses district operations, finances, transportation, food service, and administration in substantial detail, constituting significant coverage rather than a directory-style mention.
The Arizona Republic (Alltucker, Oct 2025) — covers the district's specific test-score outcomes, not merely the district's existence.
12News (Nov 2025) — covers the district's successful 15% M&O override vote with attribution from election results. This is independent local news coverage of a district-specific funding issue.
Calling all county-frame coverage "routine" reads WP:CORPDEPTH too narrowly. These are not mere listings or passing mentions; the sources address the district directly.
On the authorship speculation:WP:NGOVT was simply a mistaken shortcut on my part, acknowledged on my talk page. A typo is not evidence of LLM-generated content. Per WP:AGF and WP:ASPERSIONS, the discussion should focus on the article's sourcing and notability rather than speculation about authorship. Sparks19923 (talk) 15:44, 27 May 2026 (UTC)[reply]
Was an LLM used to generate this response? Plain clear answer please.
Audits of a school district strike me as extremely routine. Essentially all school districts do them--it's a legal requirement in many places. This does nothing to demonstrate notability.
The test score article mentions Arlington once, in one sentence. This is not substantial. In fact "mere listings or passing mentions" is an extremely good description of this source.
Delete as notability is not shown, and I don't think it can be. Also, contrary to post above, "is it LLM" is a legitimate question (otherwise our new LLM policy could never be enforced) and I think the answer is fairly clear. M kuhner (talk) 16:37, 27 May 2026 (UTC)[reply]
Delete: Fails NSCHOOL, NORG and\or GNG. Reverting to a previous (non-LLM-generated) version, even keeping current sources, does not present multiple significant coverage in reliable and independent sources.
Consistently cites sources for information that is not in them, the most likely explanation for which is WP:NOLLM violation. For example, neither of the first two refs is meaningfully about weirdcore: the first uses the term a single time and the second ref does not contain the term. Delete per WP:TNT and WP:NOLLM. ~ A412talk!17:40, 27 May 2026 (UTC)[reply]
Weak Keep / TNT: The subject of the article is a real and notable internet aesthetic, but as the article exists it's completely LLM-generated. Hopefully better sources can be found. :Miiversal (talk) 15:37, 1 June 2026 (UTC)[reply]
Page on an Italian-French scientist which fails notability and had masses of fake claims indicating AI hallucinations or worse. After checking sources, I removed 10 of 15 which either did not exist, were irrelevant or routine. These included a claim that he was nominated as a knight in both France and Italy. There were other invalid claims such as that he is a full professor. Going beyond this his h-factor is 45 with 12K cites, but he is one of many on most of them. The only partial exception are two from his PhD in 2015 & 2017 where he is first author. However, his citations/year have started to drop which is not good. Clear fail of WP:NPROF, no WP:SIGCOV after removal of inappropriate claims. Page was PROD'd by Jimjamjoe5 due to lack of academic notability, which was reverted by Cyclopia with a tag of possible vandalism, I guess the sources were not checked. Ldm1954 (talk) 12:14, 27 May 2026 (UTC)[reply]
Delete: Decent citations, but is listed along with many other co-authors in most/all of the papers in Gscholar. The AI slop is a huge red flag that this is PROMO. I don't see a pass at PROF. Oaktree b (talk) 13:10, 27 May 2026 (UTC)[reply]
I'm not convinced it's (all) AI slop. The article was created back in 2015, and most of the article's content including the claim he was knighted was already in place by August 2022 [9], before the widespread availability of AI text generation ushered in by ChatGPT in November 2022 (indeed, the knighting claim has been in the article since 2017 [10]). This may just be old-skool fake citations by one of the IP authors of the article, which is still absolutely a red flag that the article has likely been edited for promotional purposes. Hemiauchenia (talk) 22:59, 27 May 2026 (UTC)[reply]
Reading the page, I think your interventions really helped in making the tone more neutral. Maybe some AI has been used to rephrase some
sentences over the years, but the page was created as said by @Hemiauchenia a decade ago. I could help in cross-checking some information ensuring notability.
I added the reference to the “full professor habilitation” status that was claimed in the page - it is easy to verify this information on Italian Public Government open access decisions - but it was removed mentioning “Both peacock and inappropriate citation. Only academics who have a full professor position qualify”. In reality, in Europe (especially in France and Italy), the habilitation to full professor is provided after an examination - which does not mean that you need to be necessarily to work as full professor at University. It is the case of many scientists working for public facilities (e.g. synchrotrons) - they can have this “habilitation” (e.g. allowing them to independently supervise PhDs, for example) but not necessarily working for a University as full professor.
Google Scholar also presents the profile with an H-Index of 45 and a 11296 citations (which is very high in terms of performance and scientific impact, and is the strongest metrics for a scientific profile of this kind) and publications appear to be on major journals, such as Nature and JACS - however, Google Scholar sometimes “amplifies” KPIs. It will be better to cross-check this information with more reliable sources, for example with Scopus. Riceboxpoulet (talk) 05:47, 29 May 2026 (UTC)[reply]
I propose we also consider a comparative approach with other scientific profiles - considering the specificity of this field, it would help. I was looking at Sophie Carenco’s wiky page (e.g. similar employer, similar research topics…) and her metrics are much lower (e.g. ~4030 citations, 40 H-Index). We have to find a way to ensure consistency Riceboxpoulet (talk) 05:56, 29 May 2026 (UTC)[reply]
That looks remarkably like WP:JUSTAVOTE. Yes, deletion is not cleanup, but that doesn't answer the suggestion that the article is "very unencyclopedic", nor the fact that leaving it in an unsuitable state for over a decade isn't cleanup either. I am also puzzled by your mention of removing an AI tag, since that has not been put forward as a reason for deletion. Why is that relevant? JBW (talk) 00:30, 27 May 2026 (UTC)[reply]
Deltaspace42 added it to the Suspected AI-generated articles delsort list. Ca's comment seems perfectly reasonable in that context. ~ A412talk!20:32, 27 May 2026 (UTC)[reply]
Keep This article covers a reasonably important finance topic that one would certainly expect to see covered in Wikipedia. The claimed reasons for deletion seem quite weak to me. Although it has had a “Needs additional citations” flag since 2010, it is by no means an uncited article and A412 recently removed most of the uncited text. I don’t know what is meant by “very unencyclopedic and not a good fit” for Wikipedia. If there are more issues to add, they should be added. It seems to be a classic case of an article that should be improved, not deleted. John M Baker (talk) 01:02, 29 May 2026 (UTC)[reply]
Article does not meet WP:SIGCOV, WP:NGAME, and has WP:LLM concerns. Going through each citation individually in the lens of SIGCOV: citations 8 and 9 don't mention the game; citations 1, 5, 7/16, 10, 11, 12, 13, 15, and 17 are of low quality (ex. citation 5 being on WP:VG/UNREL, 10 being WP:TWITTER, and 15 being WP:PRIMARY); and citations 2, 18, and 19 mention but are not entirely about the game. This leaves only citations 3, 4, 6, and 14 to provide secondary reliable coverage focused on the game. While this meets the bare minimum of SIGCOV, some of these read as WP:ROUTINE news coverage (2 and 14 specifically) which traditionally do not count towards the notability of an article. On top of all of this, sections like "Command and armory" and "Hero roster and store", which include no citations, return as 100% AI generated on ZeroGPT. For these reasons, I propose redirecting this article to its parent company Epic Games. Johnson52419:15, 13 May 2026 (UTC)[reply]
@Zxcvbnm A most valuable response (thank you for taking the time to find more sources!) and I'll differ from my original assessment and agree that good sources do exist, in line with your comment 🙂 That being said, the vast majority of this article is still 1. cited using aforementioned poor quality sources, 2. not cited at all, and 3. written using AI, that the WP:BURDEN to clean-up the article is enough to warrant a WP:TNT redirect until this objectively-terrible article's issues can be fixed. Johnson52421:06, 13 May 2026 (UTC)[reply]
It's true the article needs massive cleanup. But, the idea that it falls under TNT is in my opinion, far fetched. If you think it's a problem, reduce it to a stub. That stub still deserves to be there though, and can clearly be improved into a high-quality article. ᴢxᴄᴠʙɴᴍ (ᴛ) 21:22, 13 May 2026 (UTC)[reply]
While an AI detector is not a valid rationale for claiming AI use, the history seems to agree with the nom's rationale. The page was originally created at User:Chasemobile/sandboxin a single edit, with the first revision even heavier on AI phrases and lighter on sourcing. The user then proceeded to make many refernce additions and other edits below 400B in diff. Sure, it's been lightly edited, but the underlying wording is all the same. Somepinkdude (talk) 22:54, 13 May 2026 (UTC)[reply]
Hi @Johnson524, I can see the issues you've raised regarding sourcing and article quality. Looking back at WP:VG/UNREL, I can see that some of the sources I used were not the best choices, even though I originally believed they were acceptable for supporting information about Battle Breakers.
I also want to clarify that I did not use AI to write the article. I tried to write in a neutral and professional style in line with Wikipedia:Neutral point of view and Wikipedia:Writing better articles, though I understand that some sections currently need significant rewriting, better sourcing, and cleanup.
Since the nomination, I have found additional reliable coverage discussing the game directly, including sources such as gamesbeat.com and a GDC presentation discussing the game's development. The issue is that most articles covering Battle Breakers only mention the game briefly and do not go very in-depth. What I wrote was based on information I gathered across the sources I linked together alongside ones that I regrettably didn't link. When planning and drafting the article in my notepad and later sandbox, I also used YouTube videos from larger and more trustworthy creators discussing the game to help gain more of a understanding about it, though I understand YouTube itself is most likely not acceptable to use as a citation on Wikipedia. Because of that, rather than linking the YouTube videos directly, I only included website sources that I used alongside them to support the paragraphs I wrote whenever possible.
I fully agree the article requires substantial improvement, especially regarding citation quality and uncited sections, however, I believe these are ultimately fixable issues better addressed through cleanup, trimming, and restructuring rather than deleting or redirecting the article entirely. Even reducing the article to a smaller stub temporarily, like what @Zxcvbnm recommended would make more sense to me than removing the article altogether. Chasemobile (talk) 23:58, 13 May 2026 (UTC)[reply]
Draftify; delete second option. The topic is certainly notable, but LLM or not, this article has substantial WP:V problems where time in draftspace would help. Large portions of the article are uncited: Gameplay has ten paragraphs without citations and development has five. ~ A412talk!22:58, 14 May 2026 (UTC)[reply]
To add, even the statements that are cited are very questionably so. Battle Breakers was developed by Epic Games and its subsidiary studio Chair Entertainment, the developer previously known for Shadow Complex and the Infinity Blade series is cited to MobyGames, an user-generated source. Chair co-founder and creative director Donald Mustard had previously helped guide Epic's transition into mobile development during the early 2010s is cited to a source that says nothing about a transition into mobile games, just that Infinity Blade was made. Archived footage, screenshots, and development assets from the unreleased title later surfaced publicly in 2023 through video game preservation communities and online archivists is cited to a pseudonymous Twitter account. ~ A412talk!23:04, 14 May 2026 (UTC)[reply]
The "pseudonymous twitter account" belongs to a known leaker who is believed to be a former Epic Games employee. The owner of the biggest Fortnite STW archival project, Fortchive had managed to find remnants of Fortnite: World Explorers in the files of Battle Breakers (shown here), which link the two together. I am currently reaching out to @MrTalida and @pox1016, who are responsible for leaking and verifying the information to get more details regarding the leak and the verification process. Chasemobile (talk) 23:35, 14 May 2026 (UTC)[reply]
WP:STUBIFY the significant offending content out and start from scratch. The identified sources make this marginally pass WP:GNG. If nobody is willing to do this, then draftify. ＶＲＸＣＥＳ (talk) 08:01, 21 May 2026 (UTC)[reply]
WP:STUBIFY as per Vrxces. I would be willing to do the stubification if that's the consensus (please ping me). The large unsourced sections, whether LLM or not, are a big problem as they stand. M kuhner (talk) 17:25, 21 May 2026 (UTC)[reply]
I missed that the original author is offering to revise. Will support draftify in that case; struck through previous !vote. Apologies. M kuhner (talk) 17:27, 21 May 2026 (UTC)[reply]
The following discussion is an archived debate of the proposed deletion of the article below. Please do not modify it. Subsequent comments should be made on the appropriate discussion page (such as the article's talk page or in a deletion review). No further edits should be made to this page.
The above discussion is preserved as an archive of the debate. Please do not modify it. Subsequent comments should be made on the appropriate discussion page (such as the article's talk page or in a deletion review). No further edits should be made to this page.
Promotional i.e. non-neutral content, unsourced false content which is not supported by historical sources at all and is most supported by local subcontinental family tradition, not encyclopedic at all, seems to be the view points of an editor who presumably is related to the subject and has expressed their own emotions when their claim has been refuted on this article itself, and details of the actual Quraysh tribe are already it's own article. — Preceding unsigned comment added by Al-Budhi (talk • contribs) 06:36, 25 May 2026 (UTC)[reply]
On one hand, Qureshi is a surname, and Wikipedia contains numerous biographies of people with the name. On the other hand, there does seem to be a fair amount of non-neutral commentary in the article history. A proper surname page is warranted, but I'll leave it to others to determine if that can be done through regular editing, or if it requires TNT. Cnilep (talk) 05:12, 26 May 2026 (UTC)[reply]
Keep: it's a surname. I've added a ref from the OUP's Dictionary of American Family Names. Note that there exists List of people with surname Qureshi. If there is very little sourced content about the name itself, it would be sensible to merge that list into this page - it's a common format, and means that the reader looking for a particular person of whom they only know the surname will have one less click to make. If there are content disputes about the article, those should be resolved on its talk page, not at AfD. PamD13:16, 26 May 2026 (UTC)[reply]
Do you agree that the list should be merged (back) into the name article, so we end up with it all at Qureshi rather than "List of ..."? So it's really "Keep and merge the list into the name page". PamD16:50, 1 June 2026 (UTC)[reply]
Per Wikipedia:WikiProject AI Cleanup/Noticeboard#Potential LLM generations by Input Zoom, these are WP:LLM articles consisting of entirely AI-generated content not verified by a human. Unlike some of Input Zoom's articles, these have not been significantly cleaned-up by other editors, meaning the issues persist. I'm requesting drafifying these articles until they can be cleaned up, or at worst a WP:TNT deletion. I also nominate the following Polish places and transportation articles with the same concerns:
Support but keep Wrocław Świebodzki–Zgorzelec railway: Johnson524, first of all, did you not bother checking the Page Statistics or even the page history to see that I have contributed to that article? So you should notify me about these nominations, or at least even just send a message on WikiProject Poland about this. To be fair you did add it to the article alerts of WP PL, but this occasion happened to be my first time checking the article alerts page. This exact article has already been previously drafted but I specifically asked it to not be drafted (1, 2) as I began adding to it, and cleaning it up. I will also go back once again to the article to add more to it, so please do not delete it. I removed reverted the notice on the article. Fortek67 (talk) 13:26, 23 May 2026 (UTC)[reply]
Comment Of course I checked. You've not edited the article for almost two months so I did not think to ping you, and while I appreciate you removing the route map and cleaning up the infobox and wikilinks, the actual prose of the article remains very similar to the AI-version created by Input Zoom, which still likely includes hallucinations. This isn't deletion, just draftication, and I still believe that remains justified until the prose can be verified. You clearly care about the article though which is what ultimately matters if the articles ever getting fixed, so I'll strike this page. Johnson52416:57, 23 May 2026 (UTC)[reply]
This article has many citations, but none of them are independent of the subject of the article. I've searched for better cites, but all I'm finding is press releases or reposts of press releases. This doesn't meet either WP:NCORP or WP:GNG and so should be deleted. MrOllie (talk) 21:43, 19 May 2026 (UTC)[reply]
Keep: Found this in the AfD queue. The sourcing concern is addressable. Education Week covered IMS Global directly and independently: Benjamin Herold reported on the OneRoster standard in January 2016 and on the Ed-Fi/IMS collaboration in February 2016; Michelle R. Davis covered districts requiring IMS certification as a procurement condition in March 2018. These are bylined news articles in a major education trade publication, not press releases. I've added those cites to the article. The organization also appears in the 1998 Wired article already cited. That's three independent sources with named authors covering the organization's standards work — enough for WP:NCORP. I'm also working on a rewrite of the prose. Sparks19923 (talk) 00:03, 21 May 2026 (UTC)[reply]
AI-generated "X in Y" article. This is a real subject, and all of the references I checked do exist. But the article is so superficial and one-sided it reads like an investment pitch. The statement that AI tools have been particularly useful for drug discovery...for tasks like virtual screening, structure-activity relationship (SAR) modeling, and de novo molecule generation is cited to a source that does not mention pharmacy or pharmaceutical science at all, and is cheerfully, confidently wrong (I have some knowledge of this field and I can confidently say AI's results in drug discovery have been an extremely mixed bag, to put it charitably). Note also the abbreviation defined once and then never used again. In principle these faults could be corrected with editing, but the article is such a mess that it needs to be deleted per WP:TNT. Let a competent human that doesn't keep Sam Altman pinups above his bed write the article. WeirdNAnnoyed (talk) 22:52, 18 May 2026 (UTC)[reply]
Keep — I have rewritten the article since my earlier Draftify comment and the position has changed. The nominator's specific concerns are now addressed. The misused citation (Sarker 2022, cited for pharmaceutical science claims despite covering only general AI) has been removed and replaced with two peer-reviewed drug discovery reviews. The unsourced claim tagged {{cn}} since June 2025 has been removed. The drug discovery section, which the nominator correctly identified as one-sided, now includes a limitations paragraph covering training data dependence and the gap between virtual screening performance and clinical trial success. Weasel constructions ("Experts say...") and promotional register have been removed throughout. The article now covers drug discovery, drug delivery, drug safety, clinical decision support, pharmacy operations, medication adherence, regulatory frameworks, and ethical challenges, with 14 validated inline citations drawn from peer-reviewed journals including Nature, Chemical Reviews, and Pharmaceutics. A pharmacy automation image has been added. The article has been assessed at B-class. The topic clearly meets WP:GNG — coverage in major pharmaceutical and clinical pharmacy journals is extensive. The original quality problems have been corrected. WP:TNT is no longer warranted. — Preceding unsigned comment added by Sparks19923 (talk • contribs)
Relisted to generate a more thorough discussion and clearer consensus. Relisting comment: Can we review the rewrite or is this unsaveable? Please add new comments below this notice. Thanks, SpartazHumbug!06:53, 26 May 2026 (UTC)[reply]
I read the revised version and while it has some decent sources, it's fairly awful. I can't quite recommend delete because of the sources, but I'm sympathetic to the idea of rewriting from scratch, because the flabby AI writing is painful. It's a grab-bag of stuff with no logical connections (how is a pillbox that records when it's opened "AI"?) and everything is so vague and insubstantial. It's all "have been proposed" / "have been applied" / "have been used". What do we learn by hearing that X has been applied to Y? What happened? Did it work?It also suffers from not ever clarifying what it means by "AI." LLMs? Expert systems? Machine learning? (Unfortunately the literature may not clarify this either. There's money to be made in being equivocal about "AI".)Also--though this could be fixed--a lot of the sources are multiple instances of the same few articles, sometimes with minor variations in citation formatting. That makes me wonder if a source to text correctness check has actually been done, because I'd think this would have been noticed. M kuhner (talk) 07:54, 27 May 2026 (UTC)[reply]
Draftify: This can be fixed, it just has to be recited and redone from scratch with no LLM usage. This article is clearly not ready for the mainspace, so it should have some time and sent to AfC when the new version is complete. ★ Campssitie (msg) (contribs) 🧋🏖09:02, 27 May 2026 (UTC)[reply]
Draftify: This is a legitimate topic, but the article is clearly not ready for mainspace. In particular, what kind of AI is being used? What happened as a result of its applications? I think we should at least keep the sources list, flawed as it is, but any version of this article suitable for mainspace would have to be a major revision of what we currently have.
I have fixed all the duplicate citations, bumping the count from 19 citations down to 13. Some of these were in the same sections and one instance was in a single paragraph. ScienceD90 (talk) 03:13, 1 June 2026 (UTC)[reply]
The following discussion is an archived debate of the proposed deletion of the article below. Please do not modify it. Subsequent comments should be made on the appropriate discussion page (such as the article's talk page or in a deletion review). No further edits should be made to this page.
Contested BLAR to Yoshua Bengio#Career and research, where this should be more appropriately covered. Fails WP:NORG: To satisfy the WP:NORG notability guideline, an organization needs to have multiple reliable sources that provide in-depth coverage of the organization, and these sources should not just regurgitate press releases, organization announcements, executive interviews, or whatever else the organization has to say about itself.
Most of the independent sources merely report on Bengio's launch of LawZero in 2025, mainly regurgitating Bengio's own thoughts about the organization without independent analysis of the organization. There are some later sources with fundraising news from various other organizations ([11], [12]), but they also mainly just repeat the organizations' announcements and quotes from executives instead of independent analysis. It is telling that these sources are not discussing the results of LawZero's research, which still appears to be in a preliminary stage.
Separately, the article appears to be LLM-assisted in violation of WP:NEWLLM (e.g. Federal AI minister Evan Solomon described the amount as "substantial" and framed the commitment as part of a broader Canadian regulatory, legislative and technical strategy for fostering trust in AI.). However, scrubbing the LLM usage will not solve the notability problem. Helpful Raccoon (talk) 01:49, 18 May 2026 (UTC)[reply]
Unless you're volunteering, draftifying might as well be the same as deletion. I see no reason to let it fester for six months. MediaKyle (talk) 10:30, 25 May 2026 (UTC)[reply]
Comment This is my assessment of the sources, excluding obviously non-independent sources such as press releases:
Extended content
Created with templates {{ORGCRIT assess table}} and {{ORGCRIT assess}} This table may not be a final or consensus view; it may summarize developing consensus, or reflect assessments of a single editor.
Reporting on Bengio's announcement of the launch of LawZero. The only coverage about the organization is derived from Bengio's announcement and his comments to the publication.
Redirect per the source assessment chart. Yoshua Bengio is a clear redirect target and draftification doesn't make much sense in this case since the subject is not notable and won't become notable through normal editing. Lovelyfurball (talk | contribs) 14:44, 29 May 2026 (UTC)[reply]
The above discussion is preserved as an archive of the debate. Please do not modify it. Subsequent comments should be made on the appropriate discussion page (such as the article's talk page or in a deletion review). No further edits should be made to this page.
Seems like a WP:LLMT of the page on fr.wiki with copied referencing code "Retrieved 2021-01-02", at best the references have not been checked to the minimum standards expected for new pages on en.wiki, at worst this is the product of an llm. JMWt (talk) 18:09, 16 May 2026 (UTC)[reply]
Keep. I reviewed this article while on new page patrol and I approved it as the siege of Bletterans did occur in 1637, and IMO, it was a notable event and the article meets WP:GNG.(2004 ref1926 ref, 1881 ref). The FR-WP article was created in 2001 (prior to the Nov 2022 introduction of ChatGPT). If this article show signs of LLM translation, that is acceptable (per the March 20th directive) (with appropriate attribution). Could this article's referencing be corrected and/or improved? Of course; it's brand new. --Rosiestep (talk) 19:39, 16 May 2026 (UTC)[reply]
no. When translating text from a non-English Wikipedia into the English Wikipedia mainspace, you may only use these tools when: You are skilled enough in both the origin language and English to confirm the translation is accurate
1. You have checked for, and removed, all AI hallucinations and other core content policy violations
2. You have checked the sources in the origin language article and you are sure the translated text reflects them fairly, and each fact or claim likely to be challenged is supported by an inline citation to a reliable source; and
3. You have complied with all the usual translation processes including terms of use-compliant attribution. from WP:LLMT
You have not checked references sufficiently if you have just copy&pasted code from another language wiki. The end. JMWt (talk) 20:21, 16 May 2026 (UTC)[reply]
that would be believable if there had been any effort made by the page creator to address this, very obvious, issue. Anyone can do this (and yes, I have also done it myself with some pages nominated in this way): you take the page back to draft, rewrite it line by line checking each reference by hand. Either someone does it now or the page is deleted until someone is prepared to do it. JMWt (talk) 08:10, 18 May 2026 (UTC)[reply]
AfD is one of the few tools available to tackle slop. This is not clean-up, this is not even about notability, it is about the fundamental basis of this encyclopedia. If AI and cut&paste is deemed a "sourcing issue" that can be retained for future editors to sort out, we might as well give up now. JMWt (talk) 15:24, 23 May 2026 (UTC)[reply]
Relisted to generate a more thorough discussion and clearer consensus. Relisting comment: NOTCLEANUP does not exempt content from NOLLM. Has the content been checked by a human? Conversely, is there evidence that no content is free of LLM issues? Please add new comments below this notice. Thanks, Vanamonde93 (talk) 22:51, 23 May 2026 (UTC)[reply]
Draftify or keep only if rewritten from checked sources. The siege appears to be a real and plausibly notable event, and sources cited above and in the article support that Bletterans was taken in 1637 during the Ten Years' War. The current article, however, has the problems identified in the nomination: it reads like a close translation of frwiki, lacks visible translation attribution, and appears to contain copied citation metadata rather than claim-by-claim source verification. That is a WP:LLMT / WP:NOLLM problem rather than a simple cleanup issue. If an editor is willing to replace the text with a human-checked version based on the cited French and English sources, I would support keeping the topic; otherwise draftification is the better outcome than retaining unverified translated text in mainspace. Sparks19923 (talk) 20:00, 24 May 2026 (UTC)[reply]
Keep. I'm fairly certain this is not an unreviewed LLM translation of the French Wikipedia article, though it would help if @Neddyseagoon could comment on this. For one, it was created in several chunks, out of order, over the course of nearly an hour and a half. (The diff sizes are messy, as the creator added the entire French text in invisible comments in their initial edit. For example, the second major content addition [13] has a negative byte change.) Then, there's the text itself:
Stuff like "La Franche-Comté est à présent quasiment sans défense." being translated as "The siege's end left the Franche-Comté almost undefended." requires more context than a straightforward translation – simply feeding that paragraph into a translation tool or LLM, it would have no idea we're talking about a siege at all. For example, Claude gives "Franche-Comté is now virtually defenceless."
"bourgeois de la ville" was translated as "middle-class inhabitants of the town", which is a rather remarkable choice that I couldn't get Claude or Gemini to spit out even after asking for alternatives.
Even the first sentence has a different structure and additional context in the English version: "à Bletterans et dans ses environs, durant l'épisode comtois de la guerre de Trente Ans." vs. "in and around Bletterans during the Ten Years' War, itself part of the wider Thirty Years' War."
The sentence "The Franche-Comté's army was beaten at Cornod on 13 March 1637 and by that summer it had almost ceased to exist." contains a date (13 March) not present in the frwiki original, but verified by archives of the cited source [14] (13 mars 1637 bataille du château de Cornod : l’armée comtoise forte de 3000 hommes dont 600 cavaliers et artillerie investit Cornod. ... Les Comtois surpris par l’arrivé des français et pris de court durent battre en retrait, assez rapidement, vers Arinthod en laissant une partie de l’armée. Le combat se transforma en véritable massacre, près de 600 morts (même 1200 si l’on se fie à la gazette de France de l’époque) et 400 prisonniers.). (Whether "almost ceased to exist" is justified is a matter of interpretation.)
I've stopped there since I feel that's enough evidence, though I'm sure looking through the rest of the article would turn up additional examples. Finally, the assertion that no attribution was provided is contradicted by the very first edit summary [15], which links back to the frwiki article. Toadspike[Talk]10:06, 25 May 2026 (UTC)[reply]
I checked some of these with Google Translate as well (without giving it any additional context);
bourgeois de la ville → "townspeople"
La Franche-Comté est à présent quasiment sans défense → "Franche-Comté is now virtually defenseless."
à Bletterans et dans ses environs, durant l'épisode comtois de la guerre de Trente Ans. → "In Bletterans and its surroundings, during the Franche-Comté episode of the Thirty Years' War."
I don't know what we are discussing now, to be honest. I asked Claude to translate the whole page and it spat this out (current text from page first, Claude output in green)
"The Franche-Comté's army was beaten at Cornod on 13 March 1637 and by that summer it had almost ceased to exist" By the summer of 1637, the Comtois army, having been defeated at Cornod, had virtually ceased to exist
"during the Ten Years' War, itself part of the wider Thirty Years' War" during the Franche-Comté episode of the Thirty Years' War.
"by some middle-class inhabitants of the town" as well as several of the town's burghers
I do not see how this shows that the content on the page is or isn't machine translated. What we can say is that the page layout, the paragraph construction and even the sentence construction closely mirrors the French version. JMWt (talk) 10:07, 26 May 2026 (UTC)[reply]
Each of these LLM examples is different from the original French article in exactly the way I described. The first is missing a date, the second doesn't use the term "Ten Years' War", and the third doesn't use the term "middle-class". This does nothing to disprove my point that a human has at the very least edited the text after machine translation, if not translated the whole article manually. I am not thrilled that Neddyseagoon hasn't bothered to defend themselves here, but faced with no evidence of LLM-introduced errors and plenty of evidence of human intervention, I cannot support deletion on the grounds of LLM use alone.
I must add that I appreciate you bringing this article here for further review and discussion, regardless of whether it ends up deleted or not. Analyzing complex situations like this is exactly what AfD is intended for, and without people to flag potentially problematic articles, it cannot serve that purpose effectively. Toadspike[Talk]23:05, 27 May 2026 (UTC)[reply]
I do not see how this shows that the content on the page is or isn't machine translated.
It doesn't. What Toadspike pointed out just made me less sure of my initial assessment, and I don't feel confident enough in stating that I am sure that this is a machine translation anymore regardless of what I might think (or not think) for myself. What I can say though I that I agree with you starting this AfD (the 'D' can also stand for Discussion), the worst case scenario is that you brought some much-needed scrutiny to an article that needed it. --gurkubondinn12:54, 2 June 2026 (UTC)[reply]
Keep, per @Toadspike. I also would like to add that, if it was LLM-assisted, the piece-by-piece nature of the edit history suggests that the translator did verify the translation and/or citations as they worked. Also, @Neddyseagoon has nearly 140,000 edits, and 55 of these are tagged as translations, with the earliest translation from French being in 2006. This is clearly an experienced Wikipedian who is experienced at French-to-English translation. Just-a-can-of-beans (talk) 09:45, 31 May 2026 (UTC)[reply]
Relisted to generate a more thorough discussion and clearer consensus. Relisting comment: Merge where? Assume the closer only looks at the discussion so everything they need needs to be here Please add new comments below this notice. Thanks, SpartazHumbug!06:11, 20 May 2026 (UTC)[reply]
Has been deleted before and as far as I can tell nothing much has changed. The most reliable sources used in the article are mostly talking about how the group maybe doesn't really exist. Keyworth describes it as little more than a money-making scam. (p.366). The article is also pure LLM as far as I can tell. David Palmer//cloventt(talk)20:59, 11 May 2026 (UTC)[reply]
If there is sigcov on it, then it is notable, whether the group is actually a membership organization or a scam or not. That an organization is good or bad or legitimate or illegitimate is of no relevance to notability. LLM issues if present may warrant a deletion but just because the group isn't a "real religion" doesn't mean it should or shouldn't have an article. PARAKANYAA (talk) 23:07, 11 May 2026 (UTC)[reply]
And the last deletion was in 2006, when notability was based on vibes, and not any guideline. From a search, the subject is 100% notable, but there are signs of LLM involvement, so I am sympathetic to a deletion from that angle. PARAKANYAA (talk) 02:03, 12 May 2026 (UTC)[reply]
Keep: This Taylor&Francis source seems legitimate, as well as both book sources. Yes, it might be a scam or a hoax, but Wikipedia has articles on scams and hoaxes, as long as they are widely reported. As for AI, I can spot a few obvious AI tells, but I'm not sure the whole thing was fully AI-generated. It looks more like an AI copyedit or translation. Somepinkdude (talk) 23:23, 11 May 2026 (UTC)[reply]
Weak Delete The fact that it's a scam makes it more notable--unfortunately though I just can't find the sources. I found that Taylor and Francis article linked above--but can one scholarly source be enough to keep the page? I'm not sure. Agnieszka653 (talk) 14:54, 18 May 2026 (UTC)[reply]
@Agnieszka653 It has an entry in Melton's Encyclopedia of American Religions, sigcov in The Occult World, the Laycock book, at least, as well as lots of other coverage. I have verified at least that most of the sources in this article (except the ones that predate the group, and Williams) do discuss them.
Keep. I've stubified the article and verified the remaining content (had to remove a couple words) against the Melton's Encyclopedia of American Religions source. The subject is notable, as the sources PARAKANYAA notes show. The entry in Melton's Encyclopedia of American Religions is 434 words, for note. Katzrockso (talk) 05:02, 21 May 2026 (UTC)[reply]
This article has several signs that it is AI generated, such as the second page edit in which all substantial content is added, and the note at the bottom of the article, which says ChatGPT was used as a search engine to search for sources related to the topic.[38] All sources were read and verified by authors of this article. Only unreviewed AI generated articles meet the WP:G15 speedy deletion criteria, and as it apparently passed through the Articles for Creation process, and has seen subsequent edits. Mitchsavl (talk) 07:02, 26 May 2026 (UTC)[reply]
Delete per WP:TNT. This is a real topic, and the sampling of sources I checked are all valid (I didn't try to verify the claims they were cited for). And I appreciate the authors' honesty about using an LLM. But the vast majority of sources are primary, and the few secondary sources aren't specifically about the subject of the article (Ref. 2 comes close). The article reads like a review paper rather than a summary for a nonexpert. A simplified version of this article can be rewritten from scratch. I hope the authors learn that creating an exquisitely-detailed article on a highly technical topic, and then admitting you had AI assistance, is not a great way to build credibility. WeirdNAnnoyed (talk) 21:45, 26 May 2026 (UTC)[reply]
Keep. Although this article does state usage of an LLM, it was not used for any writing, but only for gathering sources (in addition to other methods like PubMed and google scholar), to ensure the authors covered the breadth of the topic, which does not violate wikipedia guidelines. Additionally, this disclaimer was added based on wikipedia's guideline on disclosing LLM usage, for clarity and honesty (see: Wikipedia:LLM use disclosure). Furthermore, the claim that substantial edits were made to the article in one of the edits is not valid evidence for AI-generated content, as we collaborated on a separate document prior to inclusion on wikipedia. Furthermore, sources were added later to prevent duplication during collaboration. We were also aware that the usage of AI for content generation is not allowed (see: Wikipedia:Artificial intelligence). With regards to the majority of sources being primary, we provided both primary and secondary sources to enhance the validity of claims in the article. Although this can be improved to focus on using secondary sources only, I don't think this should be a reason for deletion because this still follows the general guidelines for notability based on the Help:Your first article page, and by comparing to other related wikipedia pages, the amount of primary vs secondary sources is comparable (see: Perturb-seq). Please clarify if I have misunderstood any guidelines. — Preceding unsigned comment added by Am245871 (talk • contribs) 02:03, 27 May 2026 (UTC)[reply]
Hi @Am245871, are you a subject-matter expert in this topic, e.g. have post-graduate certifications (MSc, PhD, post-grad etc)? If so, I would recommend reading WP:EXPERT as there is some helpful advice there. I would also like to say good work so far, hopefully you will continue to contribute more articles to Wikipedia. Felinaex(purr / pawprints)18:19, 2 June 2026 (UTC)[reply]
Keep Real topic, notable, no proof that LLMs were used to generate the text (and honest acknowledgement they were used to search for sources, which is legitimate and useful if done carefully). --cyclopiaspeak!15:40, 27 May 2026 (UTC)[reply]
The substantial content added in the second edit did not add the current citations, and instead had in text attributions. The proper citations were only added afterwards, in subsequent edits, as seen in the third one. Looking at it, the citations added doo seem to correspond to the ones previously in the in text format, so the sources could have been found before and then reformatted into it.
What I found confusing to me, and raised my suspicions of AI use, was all the content being added all at once, instead of being worked on and developed from scratch in the project space. I was also thrown off by the AFC template, I didn't realise it showed as having passed through AFC even on revisions before it had, though I have checked and it appears it does work that way. This led me to jump to the conclusion that the article was created in mainspace, in a single edit, with the template added under improper use, and without following policy, instead of having gone through the correct processes as it appears to have done. Mitchsavl (talk) 05:23, 29 May 2026 (UTC)[reply]
Keep, I second User:Cyclopia's points. A quick search of Google Scholar shows numerous mentions of its use as a technology, and I appreciate the honest admission of LLM use to gather sources and the clarification that the sources had been read and verified by authors of this article. I personally don't use LLM but I disagree with the kneejerk reaction that LLM = bad. WP needs more WP:EXPERTS. Just because it's currently written like a research paper doesn't mean it can't be improved. Disagree with WP:TNT, it does need rewriting to make it more accessible to a lay reader but overall it's a good article. Felinaex(purr / pawprints)18:15, 2 June 2026 (UTC)[reply]
AI Source Verification - Userscript that uses open-source models (Free!), Claude, Gemini (Free!) or ChatGPT to help check if a source supports a claim.
CitationVerification - Python script that uses MiniCheck and Claude to check if a source supports a claim.
Shoe company Allbirds announces it will be pivoting to providing computing infrastructure for artificial intelligence, with plans to rename itself to "NewBird AI". The company's shares rose nearly 600% on the news. (CNBC)
