Google Books is indexing low quality, AI-generated books, which may impact Google Ngram viewer, an important tool used by researchers to track language use
It's almost like they need... A librarian. — https://www.404media.co/... Jason Koebler / @jasonkoebler@mastodon.social : Like Amazon, Google Books has also been infected with lots of AI-generated garbage. This is bad for a variety of reasons. One of them is that Google Books ultimately feeds into NGram Viewer, which is a tool academics use to do research on human culture and language — https://www.404media.co/... Alex Hanna / @alex@dair-community.social : Google Books is indexing a bunch of AI-generated trash, likely coming from Amazon or associated ebook marketplaces. Thanks to Emanuel and 404media for covering this. — https://www.404media.co/... Joseph Cox / @josephcox@infosec.exchange : New: ChatGPT continues to pollute spaces where people go to acquire knowledge. ChatGPT generated books are being indexed in Google Books, and could impact Google Ngram viewer, an important tool used by researchers to track language use throughout history https://www.404media.co/... [image] Sam Cole / @samleecole@journa.host : AI generated books, indexed by Google, threaten to render research tools “completely unusable” https://www.404media.co/... Bluesky: @catbus.bsky.social : *hooking the sewer pipe up to my water supply* — this is a great idea I am very intelligent [embedded post] @anubisthejackal.bsky.social : [ 20 years from now ] — “What was the internet like?” — Well, kid, before a bunch of booger-eating idiots strip mined it for money, it was a really cool interactive way that people could talk to other people, and buy stuff from far away stores. [embedded post] @obicynkenobi.bsky.social : If Google, Amazon and others don't start jumping on managing this shit more asap; their user bases will start to spend more time finding alternatives. — I already search TikTok for certain things more than Google, & I've debated looking into using Bing as my default search engine now? [embedded post] X: Ali Alkhatib / @_alialkhatib : this piece is great. something i don't see in the article is that (as far as i know) google mostly accepts reported publication dates at face value, so if someone wants to taint historical data, they could claim the book was published earlier than ~2024 https://www.404media.co/... Alex Hanna / @alexhanna : Google Books is indexing a bunch of AI-generated trash, likely coming from Amazon or associated ebook marketplaces. Thanks to @emanuelmaiberg and @404mediaco for reporting on this. Joseph Cox / @josephfcox : For example, look at this. Scrolling through Google Books, tons of clearly AI-generated garbage. ChatGPT is infecting more and more spaces that are supposed to be about specifically human knowledge and expression https://www.404media.co/... [video] Jason Koebler / @jason_koebler : Google Books is full of AI-generated garbage. This is bad for a variety of reasons. One of them is that Google Books ultimately feeds into NGram Viewer, which is a tool academics use to do research on human culture and language https://www.404media.co/... [video] Dare Obasanjo / @carnage4life : The snake is eating it's own tail. Google Books is becoming infiltrated by AI generated slop which makes certain search tools less useful. This is a similar problem to SEO content spam and AI generated books clogging up Amazon. AI is a godsend for spam https://www.404media.co/... LinkedIn: Christopher Rice : The best way to understand the biggest business and cultural impact of Generative AI is as a Kessler Syndrome for the Web. … Alex Hanna, Ph.D. : Google Books is indexing a bunch of AI-generated trash, likely coming from Amazon or associated ebook marketplaces. Thanks to Emanuel and 404 Media for covering this. … Forums: Hacker News : Google Books Is Indexing AI-Generated Garbage Msmash / Slashdot : Google Books Is Indexing AI-Generated Garbage See also Mediagazer
Context & Ripple Effects
The spread of AI-authored books had already been identified as part of a broader shift in which text generators produce more online and published material, with fewer commissions flowing to human writers earlier coverage of AI-authored books and articles. This report matters because the material is not confined to a retail marketplace: it enters a corpus used to study how language and culture change over time.
It also extends a recurring Google data-quality problem. Prior reporting found troubling material in Google's C4 training dataset Google's C4 dataset, while criticism of search had focused on spam and SEO incentives a spam-ridden search experience. Here, low-quality synthetic text risks distorting a research instrument rather than only degrading discovery.
First-order effects
- Researchers using Google Ngram Viewer may encounter frequency patterns influenced by low-quality AI-generated books, weakening confidence in results intended to describe human language use.
- Google Books and Ngram Viewer face an immediate corpus-governance problem: indexing decisions for books can now affect the reliability of an academic-facing analytical tool.
Second-order effects
- Scholars may need to validate Ngram findings against narrower or independently curated corpora, increasing the cost of longitudinal linguistic and cultural research.
- The apparent flow of AI-made books from Amazon or affiliated ebook marketplaces makes marketplace ingestion and provenance checks more consequential for downstream services that reuse book metadata or text.
Third-order effects
- If synthetic publishing continues to enter historical corpora without clear labeling or filtering, dataset curation becomes a core part of research reproducibility rather than a background infrastructure task.
- The episode illustrates generative editorial debt: systems that prioritize broad collection can accumulate cleanup and provenance obligations as synthetic supply scales.
The trend: Generative AI is turning content provenance and corpus curation into critical infrastructure for search, research, and any product that treats published text as evidence.