Language Customization and the Market for News

In large-scale experiments, we study people's news consumption on a news app, \textit{TailorMade Times}, that enables users to customize news articles in terms of their entertainment, negativity, political perspective, and complexity. As users set their preferences, the app dynamically generates and presents AI-created news content tailored in real time, while holding constant the underlying facts from a credible source. According to our framework, the app decreases search and matching frictions for consumers. Our experiments shed light on the demand for and the consequences of language customization. We document a series of key findings: First, we show that people's likelihood of downloading and using a news app significantly increases when it offers language customization. Second, using rich click data, we characterize how people customize their news in practice. Third, we study the causal effect of language customization on knowledge acquisition, issue polarization and affective polarization.