Inferring and simulating human behaviors for societal decision-making

Understanding human behaviors is crucial for high-stakes societal decisions: for example, effective pandemic response relies on understanding how infectious diseases spread through contact between individuals and how individuals change their behaviors in response to policies and disease. However, fine-grained behaviors are often difficult to observe (e.g., for cost or privacy reasons) or cannot be observed (e.g., future or counterfactual behaviors). In this talk, I’ll discuss two approaches to addressing this challenge: (1) inferring behaviors from novel data sources and (2) simulating behaviors with generative AI. In the first part, I will describe our work on inferring hourly mobility networks from aggregated location data and modeling the spread of COVID-19 over these networks to inform pandemic policies. In the second part, I will describe our recent work on simulating diverse human behaviors with generative AI models, from public opinions to social networks to mobility trajectories.

Speaker: Serina Chang, UC Berkeley