My research examines ethical decision-making and its foibles, first in humans and now in AI. My published work investigates when and why generally honest people come to see lying as acceptable, what exacerbates political polarization, and how ethical norms degrade in groups.

My current work focuses on how AI models behave in ethically significant situations and the processes that underlie this behavior. I take the same experimental approach, designing studies that pressure-test moral behavior and isolate what drives it. My ongoing projects examine how models mislead without saying anything false, how their risk preferences vary across domains, and how they persuade people in voting decisions.