Generative AI floods and dilutes the market for books

Analysis

Why This Paper Matters

This paper provides the first large-scale empirical evidence that generative AI is not just producing low-quality 'slop' that consumers ignore, but is actively reshaping a creative market through sheer volume. By analyzing 14,419 self-published genre-fiction books on Amazon from 2023 to 2026, the authors show that AI-generated books are gaining commercial traction—winning a growing share of sales and top-ranked positions—even as they dilute overall revenue per book. This matters because it challenges the assumption that AI-generated content has negligible market impact, and it directly addresses the market-effect question central to the fair use defense in copyright law.

The study is particularly timely as generative AI tools become widely accessible, enabling near-zero-cost production of book-length works. The findings have immediate implications for authors, publishers, and policymakers, suggesting that the threat from AI is not quality but scale: a flood of AI-generated books can crowd out human-authored works, especially in genres with high AI diffusion and Kindle Unlimited availability.

Technical Contributions

• Full-text AI detection pipeline: The authors developed a method to detect substantial AI text (>25%) in full-length books, applied to a large corpus of 14,419 self-published genre-fiction titles.
• Longitudinal sales analysis: Daily sales records from 2023 to 2026 were matched to AI detection results, enabling tracking of market dynamics over time.
• Genre-specific and platform-specific analysis: The study examines how AI diffusion varies by genre and how Kindle Unlimited availability moderates the impact on human-authored books.
• Distinctiveness metric: For top-selling books, the authors measure the overlap of language with existing books, finding that AI-text books draw on more distinctive language, with overlap increasing with revenue—a pattern not seen for human-authored books.

Results

• Books with >25% AI text make up a large share of the catalog but a smaller share of sales, yet their sales share grows over time.
• The number of books with observed sales in a quarter grew 19.2-fold, while quarterly revenue grew only 8.9-fold, indicating market dilution.
• Revenue per selling book fell across most genres, with books containing no AI text losing the most ground in genres with high AI diffusion.
• Among top-selling books, those with substantial AI text show higher overlap with existing books' language, and this overlap rises with revenue—a gradient absent for books with no AI text.

Significance

This research provides critical evidence for the ongoing legal and policy debates around generative AI and copyright. By quantifying the commercial scale of AI-generated books and their market effects, the paper directly informs the fair use defense, which hinges on whether AI-generated works harm the market for original works. The findings suggest that the harm is real and growing, not through quality competition but through volume-driven displacement. For AI practitioners, the study underscores the need for responsible deployment of generative models and highlights the importance of transparency (e.g., disclosure of AI content) in creative markets. The methodology also offers a template for monitoring AI's impact in other content domains, such as music, art, or journalism.