AI and the Quantity and Quality of Creative Products (www.nber.org)

A recent study by Imke Reimers and Joel Waldfogel investigates the impact of large language models (LLMs) on book production from 2022 to 2025, highlighting a significant surge in new book releases, which have tripled during this period. However, this influx raises concerns about the quality of these publications. Their findings reveal that while the overall average quality of books released during the initial AI boom is lower, the top 1,000 monthly titles in each category have shown improved quality, particularly in faster-growing genres. This suggests that LLMs are enabling a wider range of creative outputs but are also associated with a rise in less prestigious works. The implications for the AI and machine learning community are profound, as this research quantifies the phenomenon of LLMs boosting creative production while simultaneously emphasizing the need for careful evaluation of quality. Notably, the study indicates that authors entering the market post-LLM tend to produce more low-quality content, while established authors have improved their output. The research posits that, in steady-state conditions, LLM-driven book production could enhance consumer surplus in book markets by 25% to 50%, demonstrating the dual-edged nature of AI tools in creative industries.