AgentDynEx: Nudging the Mechanics and Dynamics of Multi-Agent Simulations
This addresses a problem for researchers and practitioners in multi-agent simulations by providing a method to balance mechanics and dynamics, though it appears incremental as it builds on existing simulation techniques.
The paper tackles the challenge of consistently enforcing simulation mechanics while allowing emergent dynamics in multi-agent LLM simulations, and presents AgentDynEx, which uses nudging to achieve more complex mechanics and maintain notable dynamics compared to simulations without nudging.
Multi-agent large language model simulations have the potential to model complex human behaviors and interactions. If the mechanics are set up properly, unanticipated and valuable social dynamics can surface. However, it is challenging to consistently enforce simulation mechanics while still allowing for notable and emergent dynamics. We present AgentDynEx, an AI system that helps set up simulations from user-specified mechanics and dynamics. AgentDynEx uses LLMs to guide users through a Configuration Matrix to identify core mechanics and define milestones to track dynamics. It also introduces a method called \textit{nudging}, where the system dynamically reflects on simulation progress and gently intervenes if it begins to deviate from intended outcomes. A technical evaluation found that nudging enables simulations to have more complex mechanics and maintain its notable dynamics compared to simulations without nudging. We discuss the importance of nudging as a technique for balancing mechanics and dynamics of multi-agent simulations.