Benchmarking large language model agent societies against human behavioural distributions

Raad Bin Tareaf

Populations of large language model agents are increasingly used as experimental societies. Three doubts shadow every such result: whether the agents behave like the humans they st…