Originality.ai: About 63% of Religious Books on Amazon May Be AI-Generated
AI content detection agency Originality.ai recently released a study stating that among a sample of 2,034 recently published religious and faith-related books on Amazon, 1,272, or about 63%, were determined to be "possibly AI-generated."
Among these, the categories of witchcraft, Hinduism, and Taoism had the highest proportions of AI-generated content, reaching 78%, 76%, and 74%, respectively.
Researchers also found that about 53% of verifiable factual claims in some witchcraft books identified as AI-generated may contain errors. However, the agency emphasized that AI detection results only represent the probability that the text "may have been written by AI," and are not a final conclusion. Due to potential misjudgments or contradictions among different AI detection tools, related conclusions remain somewhat controversial.
The study covered 14 categories of religion and faith, sampling book abstracts, author biographies, and text samples. AI-generated books tend to have lower average prices and fewer pages. Amazon relies on authors to self-disclose their use of AI, with no publicly enforced verification data.
In terms of market mechanisms, event-driven factors have led to a surge in low-cost content supply. Funding and traffic are directed toward cheap, quickly published AI-generated titles, benefiting bulk publishing tools and detection companies, while traditional human authors and accuracy-seeking readers face pressure. Categories like witchcraft, which have lower scientific standards, are more easily filled by AI.
ABAB AI Insight
Originality.ai has been monitoring the Amazon content ecosystem for a long time and previously found that AI books are generally shorter and cheaper. This focus on religious categories, particularly in subfields like witchcraft and "self-help + spirituality," has become a hotspot for mass AI generation due to the lack of strict factual constraints, with a factual error rate of 53%.
From a capital perspective, low-barrier publishing platforms allow anonymous or self-published authors to quickly list their works, while AI tools further compress marginal costs. Similar phenomena have appeared in science fiction and self-help books, with religious categories accelerating penetration due to rigid emotional demand. Amazon's honor system disclosure has failed to prevent large-scale influx.
The current publishing industry is transitioning from "human-led creation" to "AI-assisted/dominated low-end content," facing dual pressures of pricing and trust for traditional publishers and independent authors.
Essentially, this is a case of technological substitution. The mechanism is that standardized, low-verification-cost texts are prioritized by AI, while high-trust or high-expertise areas still retain human premiums, leading to a clear stratification in the content market.
ABAB News · Cognitive Law
- Areas with low verification costs are filled by AI first.
- Detection tools can only provide probabilities, not truths.
- Cheaper books are often written faster and contain more errors.