AI-Generated Books Gain Market Share, Diluting Revenue in Self-Publishing

A new working paper provides the first large-scale empirical evidence that generative AI is actively reshaping the market for self-published books, not through superior quality but through sheer volume and strategic positioning. The research, analyzing over 14,000 genre-fiction books on Amazon from 2023 to 2026, finds that AI-assisted titles are winning a growing commercial foothold, diluting overall revenue per book and displacing human-authored works from coveted top spots.

Key takeaways

• Books with substantial AI-generated text (over 25%) now constitute a large share of the catalog on Amazon but a smaller, yet commercially significant, share of sales.
• These AI-assisted books are capturing a growing portion of sales over time and are increasingly occupying the scarce top-ranking positions that were once dominated by books with no detected AI text.
• From 2023 to 2026, the number of books with observed quarterly sales exploded by 19.2 times, but total quarterly revenue grew only 8.9 times, indicating severe market dilution.
• Revenue per selling book declined across most genres, with fully human-authored books losing the most ground in genres with high AI adoption and where Amazon’s Kindle Unlimited subscription service is prevalent.
• Among top sellers, books with substantial AI text show higher linguistic overlap with existing copyrighted works, and this overlap increases with their revenue—a correlation not found for human-authored books.

The AI Flood in Self-Publishing

The study, titled “Generative AI floods and dilutes the market for books,” tested the common assumption that AI-generated book-length fiction is commercially insignificant “slop.” By applying full-text AI detection to 14,419 self-published books and matching them to daily sales records, the researchers found that none of these books disclosed AI use. Despite making up a smaller proportion of sales than their share of the catalog, AI-assisted books have reached a commercial scale that can no longer be ignored. Critically, their market share is growing, and they are successfully competing for the limited number of high-visibility ranking slots on platforms like Amazon, which are crucial for discovery and sales.

Market Dilution and the Copyright Question

The data reveals a clear pattern of market dilution. The staggering 19.2-fold increase in the number of books selling in a quarter, compared to an 8.9-fold increase in revenue, means the market is being flooded with new titles faster than it is generating new money. This drives down the average revenue per book. The impact is not evenly felt; human-authored books suffer the most in genres where AI diffusion is high. The research also notes a pivotal finding for copyright law: top-selling AI-assisted books draw on more distinctive language from existing copyrighted books than their human counterparts do, and this linguistic overlap correlates with their commercial success. This bears directly on legal debates about market harm and the “fair use” defense in copyright infringement cases.

Why it matters

This research moves the conversation about AI in creative industries from hypothetical fears to documented economic impact. It demonstrates that generative AI can reshape a market through scale and cost advantages rather than quality, potentially crowding out human creators and depressing incomes. The correlation between commercial success for AI books and their use of distinctive language from existing works also introduces a new dimension to copyright disputes, suggesting that AI-generated content may commercially benefit from its training data in measurable ways. For authors, platforms, and policymakers, the study provides crucial evidence that the AI flood in publishing is real, commercially active, and has concrete dilutive effects.