Title: Generative AI & Creative Goods: Market Expansion, Crowd-out, and Copyright

Abstract: We study how generative artificial intelligence (GenAI) affects creative goods markets using data from a large stock images marketplace. In December 2022, the platform announced it would allow artists to sell GenAI-produced images, subject to two conditions: all GenAI images must be labeled, and GenAI would be prohibited in certain markets. We exploit this policy variation using a difference-in-differences design. We estimate a 136% increase in image production and 47% increase in active artists, accompanied by a 15% decline in non-GenAI production and 29% decline in non-GenAI active artists. On the demand side, total sales increase by 82%, but non-GenAI sales fall by 28% and non-GenAI sales rates decline by 5%. We use image embeddings to develop novel measures of image quality, characteristics, and market composition. We find that quality increases by 10% overall and 2% for non-GenAI content, suggesting low-quality non-GenAI artists exit. Our product characteristic measures indicate that GenAI production is heavily concentrated in characteristic space, potentially limiting the value of these images to the market. Non-GenAI production responds to GenAI by differentiating into niches. Our results suggest that GenAI can lead to market expansion and quality improvements, but substantially crowds out non-GenAI production—a finding with direct implications for copyright policy and the longer-run sustainability of original creative work.

Bio: Samuel Goldberg is an assistant professor of marketing at Stanford Graduate School of Business. His work seeks to better understand the role of privacy, data, and monitoring technologies in markets. He holds a PhD from Kellogg School of Management at Northwestern University and undergraduate degrees in economics and physics from Brandeis University. Prior to his PhD, Goldberg served as an associate economist at the Federal Reserve Bank of Chicago.