内容简介：Gen AI expands content supply faster than platforms learn what users value. We model an attention platform on which user-generated content (UGC) and AI-generated content (AIGC) compete for scarce recommendation traffic. Every upload first enters a training pool helping the platform learn about its quality, before high-quality content is promoted to a spotlight pool. AIGC has a cost advantage, but UGC is more likely to be high quality. Consumers remain active on the platform when realized content quality exceeds their outside option, endogenizing attention. In equilibrium, lower AIGC costs increase entry and can raise AIGC’s market share but also congest the training pool, lower spotlight exposure and reduce attention. We then study provenance-contingent payouts, imperfect training design, and upload fees as screening instruments when creators privately observe noisy post-production signals. The central message is that Gen AI strains attention platforms not mainly by producing more content, but by overloading quality discovery.

演讲人简介：Chenyang Li is an Assistant Professor at the Hong Kong University of Science and Technology (GZ), working in the Innovation Policy and Entrepreneurship Department. His work involves both theory and empirical studies. On the theory side, he studies how different ways of communicating within groups can affect their ability to work together effectively. On the empirical side, he looks at how businesses work online, how they set prices on the internet, how online and traditional businesses interact with each other and adoption and effect of AI. He is dedicated to discovering new things that can help both scholars and businesses.