Generative AI is producing books at scale, flooding and diluting the market. This trend may impact authors, publishers, and readers, raising concerns about content quality and originality.
SynthePulse Insight · AI deep reading
Version 1 · 1 source
A study of 14,419 self-published genre novels on Amazon shows that AI-generated content is diluting the book market through sheer quantity rather than quality.
Generative AI can create full-length novels at near-zero cost, and these works are often dismissed as low-quality 'slop' with little commercial impact. However, a study of self-published genre novels on Amazon challenges this assumption.
The research team applied full-text AI detection to 14,419 self-published genre novels sold on Amazon between 2023 and 2026, and matched them with daily sales records through June 2026. None of the books disclosed whether they contained AI-generated content.
The results show that books with more than 25% AI-detected text occupy a large share of the catalog but a small share of sales. Nevertheless, they achieve commercial scale and gain more sales share over time, gradually taking over scarce top rankings previously held by books without AI text.
During the study period, the number of books with quarterly sales grew 19.2-fold, while quarterly revenue grew only 8.9-fold. This means the market added new selling books faster than revenue growth, leading to a decline in average revenue per book in most genres.
In genres with high AI diffusion, books without AI text lost the most, especially in areas with high Kindle Unlimited availability. This suggests that AI-generated content is diluting the market rather than expanding the overall pie.
This phenomenon reveals how generative AI reshapes creative markets through scale rather than quality, contrasting sharply with traditional quality-based competition.
Among bestsellers, books with substantial AI text tend to use more distinctive language from existing books than those without AI text. This language overlap increases with revenue, a gradient not observed in books without AI text.
This finding suggests that AI-generated content may attract readers by mimicking the language patterns of existing works rather than offering original content. This could have significant implications for fair use defenses in copyright law.
The study authors note that these results are directly relevant to the market impact question in copyright infringement fair use defenses, providing empirical evidence for policy-making.
The study is a preprint under review, and its conclusions have not yet been peer-reviewed. AI detection methods may have errors, and the sample is limited to self-published genre novels, which may not represent the entire book market.
The study does not disclose the specific algorithms and thresholds used for AI detection, which may limit reproducibility. Additionally, sales data only go through June 2026 and do not cover more recent market changes.
Although the study reveals correlations, causality remains unclear. Whether AI-generated content directly causes revenue declines or whether other market factors are at play requires further research.
This report is based on an arXiv preprint paper that is under review and has not been peer-reviewed. The research data comes from self-published genre novels on Amazon and may not represent the entire book market. The accuracy of the AI detection method has not been publicly verified, so the findings should be considered preliminary.
Generative AI is diluting the book market by mass-producing low-quality content rather than competing on quality. This trend poses new challenges for copyright law and market structure, but the findings require further validation.
Primary report
Primary source