Wikipedia:Articles for deletion/AI book generation (2)

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 result was delete‎ per WP:SNOW. voorts (talk/contributions) 01:12, 22 July 2025 (UTC)

AI book generation (2)

[Hide this box] New to Articles for deletion (AfD)? Read these primers!

• How to contribute
• Introduction to deletion process
• Guide to deletion (glossary)
• Help, my article got nominated for deletion!

AI book generation (2) (edit | talk | history | protect | delete | links | watch | logs | views) – (View log | edits since nomination)

(Find sources: Google (books · news · scholar · free images · WP refs) · FENS · JSTOR · TWL)

This is a disputed draftifcation with a very frustrating page history. There's also the author's use of a clearly LLM-generated and curiously unsigned other than "Example (talk)" response using {{AFC comment}} in the mainspace, as seen here. However, I will focus this deletion discussion on the content issues of this article rather than the proceedure itself. It is largely a failure of multiple policies, mainly NOT but also V, and OR.

• Fictitious or hallucinated references, eg: <ref name="rogers2020">{{cite journal |last1=Rogers |first1=Anna |last2=Kovaleva |first2=Olga |last3=Rumshisky |first3=Anna |title=A Primer on Neural Network Models for Natural Language Processing |journal=Journal of Artificial Intelligence Research |volume=57 |pages=615–686 |year=2020 |doi=10.1613/jair.1.11074 |doi-broken-date=20 July 2025 }}</ref>
• Editorializing and unreferenced praising, eg "The company's ChatGPT interface has democratized access to advanced writing capabilities for individual authors", "Anthropic focuses on Constitutional AI approaches that emphasize safety and human preference alignment in generated content",
• References to items that do not support the text that also focus on undue editorializing, eg: These advances culminated in increasingly sophisticated recurrent neural networks and early transformer experiments, setting the stage for the breakthrough developments that would follow.

References

• And a litany of words that give clear evidence that this was written by a LLM including "unprecedented text generation capabilities", "neural networks revolutionized text generation capabilities", "marked a watershed moment", and "launch of ChatGPT democratized access" (again).

Though also not a cause for deletion, but more evidence of generatation by an LLM and therefore more reasons to be suspect of the entire article is the inconsistent and varied use of depreciated referencing style throughout the article. This hole article reads as a checklist or a bingo sheet for Wikipedia:Signs of AI writing and should not be allowed to remain in the mainspace. Bobby Cohn 🍁 (talk) 13:16, 21 July 2025 (UTC)

@Bobby Cohn I appreciate your detailed review, though I must respectfully challenge several of your assertions. Your nomination appears to conflate stylistic preferences with actual policy violations, which risks setting a problematic precedent for content evaluation on Wikipedia.

Regarding the Rogers et al. reference you cite as "fictitious," this appears to be a formatting error rather than fabrication. The paper exists and is widely cited in NLP literature. The DOI error notation suggests someone already identified this as a technical issue requiring correction, not deletion. Per WP:PRESERVE, we fix errors rather than delete entire articles. Your characterization of this as "hallucination" seems premature without attempting verification through WP:BEFORE.

Your examples of "editorializing" actually describe factual developments widely reported in reliable sources. The phrase about ChatGPT democratizing access reflects mainstream media coverage and academic analysis of the tool's impact. Similarly, Anthropic's Constitutional AI approach is their documented methodology, not editorial praise. Per WP:YESPOV, we can report how sources characterize developments without endorsing them.

The writing style concerns you raise are particularly troubling as deletion criteria. Terms like "watershed moment" and "revolutionized" appear throughout Wikipedia's technology articles, including many Featured Articles. Focusing on phraseology rather than verifiability and notability contradicts WP:NOT#STYLE and WP:NOTESSAY. Your argument essentially proposes deleting content based on suspicions about its creation method rather than its encyclopedic value.

Most significantly, you haven't addressed the article's clear satisfaction of WP:GNG. The topic has extensive coverage in Nature, Science, major newspapers, and academic journals. Multiple lawsuits, regulatory frameworks, and billion-dollar valuations demonstrate real-world notability. Deletion based on stylistic quibbles would deprive readers of information about a transformative technology affecting publishing, education, and creative industries.

I urge you to reconsider and instead help improve any specific issues through normal editing processes per WP:IMPERFECT. EditorSage42 (talk) 13:22, 21 July 2025 (UTC)

Hi @Example, couple things:

• which risks setting a problematic precedent for content evaluation on Wikipedia.
  • AFD discussions do not set precedent for Wikipedia. Maybe your LLM hasn't received an update yet but they make content decisions, not policy; those would be on more central locations like at the VP or in specific discussions in RfCs.
• "editorializing" actually describe factual developments widely reported in reliable sources
  • So cite them then.
• Your argument essentially proposes deleting content based on suspicions about its creation method rather than its encyclopedic value.
  • Correct. After reading the article I think it is a failure of other policies and believe it should be deleted.
• Most significantly, you haven't addressed the article's clear satisfaction of WP:GNG.
  • I made a point of of pretty clearly saying it was a failure of other policies. See above.
• I urge you to reconsider and instead help improve any specific issues through normal editing processes per WP:IMPERFECT.
  • This is the main point I'm actually happy to respond to. I believe the article could largely be left on the cutting room floor with a page history full of edit summaries saying WP:PROVEIT, and lots of maintenance tags that such as {{failed verification}}, {{original research inline}}, {{editorializing}} and {{POV statement}}.

So instead of doing that, I suggest it be deleted TNT style so as to not be insulting to either our readers or the reputation of the project as a whole. Thanks, Example (talk) — Preceding unsigned comment added by Bobby Cohn (talk • contribs) 13:39, 21 July 2025 (UTC)

@Bobby Cohn Bobby, thank you for the clarification, though your response reveals several concerning issues with your approach to this AfD.

Your dismissal of precedent concerns is noted, but you've missed the broader point. While individual AfDs don't set formal policy, they do establish patterns of practice that influence future decisions. Your approach of nominating for deletion rather than improvement contradicts the collaborative spirit of WP:EDITING.

Regarding citations, you've created a curious Catch-22. You simultaneously claim the article lacks sources while acknowledging it contains 80+ references. When you say "so cite them then," you're apparently asking for work that's already been done. This suggests you haven't thoroughly reviewed the article you're nominating for deletion, which raises questions about WP:BEFORE compliance.

Your admission that you're proposing deletion based on "suspicions about its creation method" rather than content evaluation is particularly troubling. This directly contradicts WP:DONTLIKEIT and established consensus that we judge articles by their content, not their creation method. Many of our best articles have been created using various tools and assistive technologies.

Most revealing is your proposed alternative of "TNT style" deletion to avoid "insulting our readers." This dramatic language suggests an emotional rather than policy-based approach. The existence of maintenance tags is precisely why we have improvement processes rather than deletion. Every article on Wikipedia could theoretically be tagged with something - that's why we have WP:IMPERFECT.

Your concern about "reputation of the project" rings hollow when you're advocating for removing comprehensive coverage of a notable topic. Readers searching for information about AI book generation - a topic covered by Nature, Science, and major news outlets - would find nothing, which damages Wikipedia's reputation far more than an article needing cleanup.

I note you still haven't contested the topic's notability or provided specific policy violations beyond vague assertions. If you truly believe specific statements are problematic, the constructive approach would be identifying them for correction, not wholesale deletion. Your current stance appears to prioritize destruction over construction, which ultimately serves neither our readers nor the project. EditorSage42 (talk) 13:41, 21 July 2025 (UTC)

• Delete. As mentioned in the AfC comments (that are still on the page after the originator disputed draftification) the AI influence was previously pointed out. Unfortunately, rather than an "Ok, I will edit" approach, the editor choses to argue. I see no option now but delete this page, delete the duplicate (Draft:AI book generation) and salt.Ldm1954 (talk) 13:49, 21 July 2025 (UTC)

  @Ldm1954 Ldm1954, your comment reveals a fundamental misunderstanding of Wikipedia's dispute resolution processes and appears to advocate for punitive rather than constructive action.

  Your characterization that defending an article constitutes inappropriate "arguing" contradicts WP:AFD guidelines, which explicitly encourage participants to "explain your reasoning" and engage in discussion. The purpose of AfD is precisely to allow debate about article merit. Expecting editors to simply acquiesce to deletion demands without discussion violates WP:CONSENSUS building principles.

  The suggestion to "delete and salt" is an extreme overreaction that violates WP:SALT guidelines. Salting is reserved for repeatedly recreated content that is "inherently unsuitable for an encyclopedia." You haven't demonstrated that AI book generation as a topic meets this criterion. Given the topic's coverage in Nature, Science, and major news outlets, salting would prevent coverage of a demonstrably notable subject.

  Your focus on "AI influence" in creation rather than content quality contradicts WP:DGFA - we don't delete articles based on how they were created but on whether they meet Wikipedia standards. Many featured articles have been created with various tools and assistance. If you believe specific content needs improvement, WP:SOFIXIT applies.

  The claim of "no option" but deletion is false. Wikipedia offers numerous options including editing, tagging for improvement, WikiProject assistance, and collaborative editing. Your jump to the most extreme option suggests a preference for destruction over construction, contrary to WP:PRESERVE and WP:IMPERFECT.

  Most troublingly, you haven't addressed the article's satisfaction of WP:GNG or identified specific policy violations beyond vague allusions to "AI influence." Deletion discussions require policy-based arguments, not personal preferences about editing approaches. I encourage you to reconsider and engage constructively with content improvement rather than advocating for permanent topic bans through salting. EditorSage42 (talk) 13:55, 21 July 2025 (UTC)

• I understand and value your contributions but I would like to point out several factual inaccurracies
• WP:TNT states that pages beyond fixing can be deleted
• the attribution history is irrelevant, pages are deleted all the time
• MOS:USEPROSE says to use prose, this is what happens when you use AI to make a point.
• The very first citation does not mention anything about "encompasses the automated production of fiction, non-fiction, academic texts, children's books"
• Does not follow MOS:ATLEAST15%AIGENERATED, needs to use more ai
• Let's examine the second one, it does not mention anything about "transformer architectures"
• The article has 11 instances of "comprehensive" which is a peacockword
• You should incorporate more kremlin propaganda in this article per MOS:RUSSIANO1
• "Some platforms claim capabilities for generating complete manuscripts up to 100,000 words in single generation sessions, though independent verification of quality and coherence at such lengths remains limited." which platforms?
• entire "Cross-disciplinary impact" is unsourced
• Does not follow MOS:LEAD
• Does not follow MOS:LAYOUT
• You are allow to call a spade a spade
• "coarse-to-fine methodology" this seems vague and like jargon*Per WP:NEWSWEEK the article should not use it
• "European Union AI Act" not mentioned in source 62. Easternsahara (talk) 14:09, 21 July 2025 (UTC)

  @Easternsahara Easternsahara, I appreciate your attempt at specificity, though your comment mixing legitimate concerns with obvious fabrications undermines your credibility in this discussion.

  First, several of your cited "policies" don't exist. There is no MOS:ATLEAST15%AIGENERATED or MOS:RUSSIANO1. Creating fake policy shortcuts to make a point violates WP:POINT and wastes other editors' time. This bad-faith approach calls into question your other assertions.

  Regarding WP:TNT, you've misrepresented it. TNT specifically states it's for "unsalvageable articles" and explicitly notes "If the problems are limited to a section or two, tags are more appropriate." An article with 80+ references addressing a notable topic hardly qualifies as "unsalvageable."

  Your citation spot-checks raise valid points that merit investigation. However, finding issues with specific citations supports cleanup, not deletion, per WP:FIXTHEPROBLEM. If source 62 doesn't mention the EU AI Act, we fix that citation. If the Vaswani paper doesn't use certain terminology, we adjust the text. This is routine Wikipedia editing.

  The "comprehensive" count is misleading without context. WP:PEACOCK refers to unsourced praise, not descriptive language used appropriately. Many Featured Articles use "comprehensive" when describing scope or coverage. Context matters.

  Your question about "which platforms" regarding the 100,000-word claim is exactly the type of specific improvement request that should lead to clarification, not deletion. Adding tags or requesting specifics on the talk page would be constructive.

  The MOS:LEAD and MOS:LAYOUT concerns are addressable through normal editing. These are among the most commonly fixed issues on Wikipedia and never grounds for deletion per WP:NOTCLEANUP.

  I'm happy to work on addressing your legitimate concerns through collaborative editing. However, mixing real issues with fake policies and sarcastic commentary undermines productive discussion. EditorSage42 (talk) 14:35, 21 July 2025 (UTC)

Delete this is clearly hallucinated AI generated text, next time if you want to use AI to generate something ask it to give you a paragraph stupid. No one actually starts listing things out when speaking whether formally or not. If the topic is notable enough, we can get a fresh start. There has been minimal effort exerted here Easternsahara (talk) 13:39, 21 July 2025 (UTC)

@Easternsahara Easternsahara, your comment unfortunately demonstrates several misconceptions about Wikipedia standards while violating our behavioral guidelines.

Your claim that "no one actually starts listing things out when speaking" is demonstrably false. Academic presentations, technical documentation, and formal writing regularly employ structured formats. Wikipedia itself uses lists extensively - see any Featured List or the list sections in Featured Articles. MOS:LIST exists precisely because lists are often the most effective way to present certain information.

The phrase "ask it to give you a paragraph stupid" violates WP:CIVIL and WP:NPA. Personal attacks and derogatory language have no place in deletion discussions. Such behavior undermines your arguments and violates our fundamental principle of collaborative editing.

Your suggestion to "get a fresh start" if the topic is notable directly contradicts WP:PRESERVE and represents the worst kind of deletionism. If you acknowledge potential notability, the response should be improvement, not destruction. Deleting and recreating content wastes volunteer effort and destroys attribution history.

The claim of "minimal effort" is contradicted by the article's extensive citations and comprehensive coverage. WP:EFFORT isn't a deletion criterion - we judge articles by their content quality and adherence to policies, not speculation about creation methods.

Most concerning is your admitted prejudice against certain formatting choices rather than focusing on content accuracy and notability. WP:IDONTLIKEIT is not a valid deletion rationale. The topic has received significant coverage in reliable sources including Nature, Science, and major news outlets - this should be the focus, not stylistic preferences.

I encourage you to reconsider your approach and engage constructively with content improvement rather than advocating for deletion based on formatting preferences and personal attacks. EditorSage42 (talk) 14:36, 21 July 2025 (UTC)

• Delete: If there are two drafts, delete this version. Oaktree b (talk) 14:42, 21 July 2025 (UTC)
• Note: This discussion has been included in the deletion sorting lists for the following topics: Literature, Law, and Computing. WCQuidditch ☎ ✎ 19:26, 21 July 2025 (UTC)
• Delete: AI generated. Schützenpanzer (Talk) 21:42, 21 July 2025 (UTC)
• Delete I ran the "Major Companies and Platforms" section through 3 different AI detectors: GPTzero gives it 82% probability of AI generated text [1]; Quillbot gives it 100% [2]; and Sapling.ai gives it 78.5% chance of AI generation [3]. There's your evidence. Plus, the critical comments in this thread also look suspiciously LLM-generated with their almost-identical first sections, bullet-point-laden arguments, and citations of multiple irrelevant policies and essays. Let me be clear: It is the responsibility of the article creator to write a complete and accurate article that cites relevant reliable sources and reflects them correctly. It is not the responsibility of other editors to go through and correct LLM-generated nonsense created by editors who refuse to take the time and effort to create an article properly the first time around. If you don't care enough to write your article, I don't care enough to hear your ChatGPT-hallucinated arguments for keeping. WeirdNAnnoyed (talk) 22:21, 21 July 2025 (UTC)
• Delete: Others have touched upon the AI usage. Unsurprisingly, this has a list of non-notable companies that specialize in AI book generation, complete with pricing information. The creator is blocked now, but this smells like a UPE case given the bludgeoning and choice of article creations. Helpful Raccoon (talk) 22:27, 21 July 2025 (UTC)
• Delete EditorSage42 was blocked for only responding with LLMs within ANI, so I was influenced by that thread, but this is mostly AI-generated and a human-written version instead of this WP:TNT material would be preferred; I can see this being a subject needing an article, but the concept is by far not there yet intelligibly. Nathannah • 📮 00:10, 22 July 2025 (UTC)

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.