What People Almost Did: Evaluating LLM Social Simulations Beyond Behavioral Fit
LLM-based social simulations are primarily evaluated for behavioral fit, testing whether agents reproduce the actions or response distributions of the people they are simulating. However, the promise of simulation extends beyond behavioral fit. Simulations can explain human behavior, diagnose barriers, and compare larg...
Jaewon Kim, Angie Boggust
· 0 citations