77%. That's the estimated share of new "Success" self-help books on Amazon that appear to have been written by AI.
An analysis of Amazon's self-help catalog identified patterns consistent with AI-generated text across a large majority of new releases in the success and motivation subcategory. Two authors stood out as extreme cases. Noah Felix Bennett published 74 books in mid-2025 alone, at a rate of more than one book per day. Richard Trillion Mantey, with hundreds of books in his catalog, was assessed to have used AI for every single one.
One Book Per Day
One book per day is physically impossible with traditional writing. A professional writer producing 2,000 words a day - a solid sustained output - needs three to four months to finish a standard 200-250 page book. Bennett's pace implies a new book every 18 to 24 hours. The only way to hit that number is a near-fully automated publishing pipeline: feed a title and rough outline into an AI writing tool, generate chapters, do minimal editing, format for Kindle, and publish.
The economics work. Self-help ebooks sell for $5 to $15 on Amazon. Through Kindle Unlimited, authors earn per page read rather than per sale. If 74 books each generate $200 in their first year from page reads, that's nearly $15,000 from a few months of low-cost content production. The marginal cost of book number 74 is essentially nothing once the pipeline is built.
A Category Amazon Can't Filter
The problem isn't simply that AI-generated books exist. It's that they're undetectable to most shoppers. Amazon's bestseller lists, review systems, and recommendation algorithms weren't designed to identify synthetic content. A book with a real-sounding author name, a polished cover, and a cluster of reviews looks identical to a book someone actually wrote.
For readers, the practical effect is a degraded signal-to-noise ratio in a category that already had a low bar. Self-help books have always recycled ideas, but older entries at least reflected someone's genuine thinking, however derivative. AI-generated versions average across millions of similar texts and repackage the result with a new title - technically coherent prose with nothing distinctive to say.
Amazon added disclosure requirements for AI-generated content in 2023, but those rely on authors self-reporting. The analysis findings suggest that system isn't working. For anyone buying in this category, authors with a verifiable public presence and books that predate 2023 are the safer bets. Everything else warrants a closer look before you spend money on it.