AgentDynEx is designed to set up, track and repair multi-agent simulations
An ACM conference paper describes AgentDynEx as an AI system for multi-agent language-model simulations, where enforcing the rules while allowing unexpected social dynamics is a challenge.
TLDR
A September 28 post says Tao was giving a talk on AgentDynEx on Jenny Ma’s behalf at HCOMP2026/CI2026. The linked ACM conference paper presents the system as a way to set up, track and repair multi-agent language-model simulations. It says enforcing simulation mechanics while allowing emergent dynamics is difficult.
Combined views
463
1 Source, first seen 5h ago
