The field of simulating human behavior is moving towards utilizing large language models (LLMs) to drive agent interactions in complex social networks. Researchers are exploring the potential of LLM-powered agents to simulate nuanced human behavior, taking into account personal preferences, traits, and connections. This approach enables the simulation of large populations and the study of emergent phenomena, providing a complementary path to understanding collective intelligence. Noteworthy papers include:

• TinyTroupe, a simulation toolkit that enables detailed persona definitions and programmatic control via LLM-driven mechanisms.
• Large Population Models, which extend traditional modeling approaches through computational methods, mathematical frameworks, and privacy-preserving communication protocols to simulate entire populations with realistic behaviors and interactions.