LLM Agents for Modeling Group Dynamics

Simulating group dynamics and human behavior is an important problem for economists, educators, social scientists and many other specialties. Large language models (LLMs) provide a powerful new tool for getting computers to mimic human-like behavior. See how researchers build agents and environments so that LLM-powered agents can act in situations of interest. Then, see how the behavior of these agents can be analyzed and compared to human group dynamics.

In this colloquium, you will see agent-based simulations applied to the study of many kinds of groups, from classrooms to international relations. You will also see some general frameworks for creating LLM-based agents. Works presented in this colloquium are conducted in a variety of programming languages, any of which are compatible with Wolfram Notebooks.