Highlighted Publications




 

Cheng et al. (2026) ask how common social sycophancy is in leading AI systems and how it affects users’ judgments, prosocial intentions, trust, and future use. They analyzed 11 large language models across 11,587 advice, interpersonal-conflict, and harmful-action prompts, then conducted three preregistered experiments with 2,405 U.S. participants. They found that AI affirmed users’ actions 49% more often than humans. Sycophantic responses increased participants’ belief that t

Layton et al. (2026) examine whether competition among private Medicare Advantage plans can successfully replace competition from traditional Medicare in delivering higher-value coverage. They analyze national Medicare enrollment, switching, mortality, payment, and market concentration data from 2007–2022, supplemented by prior empirical research. They find that Medicare Advantage enrollment increased from about 20% to more than 50% of beneficiaries, while four national insur

Abrams and Goonetilleke (2026) examined how stop-and-frisk practices vary across major U.S. cities and whether crime, poverty, demographics, or police resources explain those differences. They analyzed more than eight million pedestrian and vehicle stops recorded in 16 cities from 2019 through 2023. They found that stop and frisk rates differed by nearly two orders of magnitude across cities. Contraband discovery rates also varied substantially, ranging from 18% to 74% for pe

Dillon et al. (2026) examined whether providing knowledge workers with an integrated generative AI tool changed how they spent their time and completed their work. They analyzed telemetry data from a randomized field experiment involving 7,137 employees across 66 large firms who were assigned either access to Microsoft 365 Copilot or a control group over six months. They found that workers who regularly used Copilot spent about two fewer hours per week on email (a 17% reducti