A lemons model of online network formation with AI-generated information

This note develops a lemons model of AI-generated information in online social networks, where individuals form provider links to receive information whose content type is hidden. Providers produce apparent information to attract subscribers, but subscribers cannot observe whether the content is human-generated or AI-generated. Hidden content type creates quality risk, so risk-averse subscribers diversify links across providers. In large markets, this attention dispersion implies that the law of the few, i.e. that only a few individuals become information providers, does not hold, and each provider becomes negligible for aggregate beliefs. As a result, in any non-collapse large-market limit, active providers become AI agent providers, with content production fully delegated to AI.