Agent-based modeling (ABM) can simulate millions of individuals and their interactions, which is crucial for policymaking. However, traditional ABMs rely on static priors, limiting their adaptability to real-time changes. Our research introduces a novel approach to bridge this information gap. Large language models (LLMs) present new opportunities for predicting human decision-making. Here, we introduce a scalable Hybrid Agent-based and Language-driven Epidemic (HALE) modeling framework that leverages LLMs to predict human decisions in ABM simulations. As a proof-of-concept, we utilize HALE to simulate COVID-19 and its impacts in Salt Lake County, UT.
Blogger's Review: This research highlights the potential of LLMs in agent-based modeling, particularly in dynamic decision-making scenarios. The HALE framework not only enhances model adaptability but also provides more accurate decision support for policymaking, making it a topic worth further exploration.