AI text generators are quietly authoring more of the internet; more AI-generated books and personalized articles mean fewer clients buying human-written content
From recipes to product reviews to how-to books, artificial intelligence text generators are quietly authoring more and more of the internet. LinkedIn: Theo Priestley . Mastodon: @willoremus@mastodon.social . Tweets: @thelincoln , @cabitzaf , @j2bryson , and @amyklobuchar See also Mediagazer LinkedIn: Theo Priestley : Makes me wonder whether human creativity will again command a premium and be preferred and #generativeai will be the cheap imitation once we hit saturation point. … Mastodon: Will Oremus / @willoremus@mastodon.social : I went on a quest to find out if AI is already secretly writing a lot of the stuff we read online. I found more than I ever imagined. — From online recipes to mattress reviews to celebrity horoscopes, human writing is fast becoming the exception on the internet, rather than the rule. … Tweets: Lincoln Michel / @thelincoln : https://www.washingtonpost.com/ ... And yeah a lot of it won't be read by humans at all https://twitter.com/... Federico Cabitza / @cabitzaf : I agree. Along with the risk of pervasive falsehood and manipulation, through (or by?) AI-generated content, there is this degradation of the mechanisms for filtering and promoting quality and for preserving variety. When too many voices, none, or only the loudest. Divide et... https://twitter.com/... Joanna J Bryson / @j2bryson : I'm not TOO worried about bad AI-generated writing making us stupider. I'm more worried about losing new voices because no on reads stuff from people they've never heard of. (gift artcl) #AIEthics He wrote a book Then a ChatGPT replica appeared on Amazon. https://www.washingtonpost.com/ ... Amy Klobuchar / @amyklobuchar : Among many issues that must be immediately addressed with A.I.: compensation for content...news (see my bipartisan bill); books; movies, and just about everything else. 👀 ⬇️ He wrote a book on a rare subject. Then a ChatGPT replica appeared on Amazon https://www.washingtonpost.com/ ... See also Mediagazer
Context & Ripple Effects
The Washington Post's reporting lands at the start of an arc this page has tracked since: writers testing whether the tools can actually write. A New Yorker writer found that generating text in his own style with Writer's LLM produced insights that felt hollow or approximated, while another writer weighed what Sudowrite's novel generator means for literature after an AI-assisted essay went viral.
What changed is scale and invisibility — recipes, product reviews, and how-to books are now routinely machine-authored without disclosure, which is exactly why Senator Amy Klobuchar has pushed for immediate action on compensating news, books, and other works in the era of AI. The economics matter as much as the ethics: every AI-generated book or personalized article is a client that never hires a human writer.
First-order effects
- Freelance writers and content shops lose paying clients directly, since recipes, reviews, and how-to books are categories where AI output is already substitutable and cheaper.
- Amazon and other book and review platforms face rising volumes of undisclosed AI-generated titles and reviews, straining whatever filtering and quality-promotion mechanisms they run.
Second-order effects
- Klobuchar's compensation push gains urgency because the displaced income is diffuse — spread across thousands of small clients rather than concentrated industries — making legislative remedies harder to target than in news licensing fights.
- Discovery becomes the battleground: as the corpus of machine-written text grows, readers and platforms must pay for provenance signals, echoing the concern that unfamiliar human voices get silenced when engagement defaults to volume.
Third-order effects
- If the pattern holds, written content splits into a cheap synthetic bulk tier and a verified-human premium tier — the outcome Theo Priestley raises when he asks whether human creativity will again command a premium once saturation hits.
- The boundary itself starts eroding: the later finding that humans increasingly mimic AI prose quirks rooted in model overfitting means even 'human' writing becomes stylistically indistinguishable, complicating any labeling or compensation regime.
The trend: Online text supply is shifting from commissioned human labor to low-cost synthetic generation, pushing the writing market toward a two-tier structure where verification of human authorship becomes the scarce commodity.