Studying the Behavior of Generative AI-Based Agents in Multi-Agent Systems

Explore a cutting-edge presentation on generative AI-based agents in multi-agent systems delivered by Joel Leibo from DeepMind at IPAM's Analyzing High-dimensional Traces of Intelligent Behavior Workshop. Dive into the world of Generative Agent-Based Models (GABMs) and discover how they leverage Large Language Models (LLMs) to revolutionize agent-based modeling. Learn about Concordia, a library designed to facilitate the creation and manipulation of GABMs, enabling language-mediated simulations in physical and digital environments. Understand the role of the Game Master (GM) agent in simulating environments and managing agent interactions. Gain insights into the potential applications of this technology in scientific research and the evaluation of digital services through user simulation and synthetic data generation.