Does audience engagement differ when content is AI-generated versus human-created, and does that gap hold up once accounting for how engagement spreads through a social network? This project combines two complementary methods: an agent-based simulation that models how engagement propagates through a stylized social graph, and a randomized field experiment that tests real engagement with human vs. AI content on Instagram.

Pairing simulation with a live field experiment lets the project separate two effects that are easy to conflate in observational data — differences in raw content quality/appeal, and differences in how that content diffuses through a network once posted.