Simulation-based frameworks can enhance the reasoning abilities of large language models in complex human systems.
