A Large-Scale Simulation on Large Language Models for Decision-Making in Political Science

Abstract

This paper presents a large-scale simulation study of large language models for decision-making in political science. Through extensive experiments, we analyze how LLMs simulate voter decisions and electoral outcomes, and investigate the factors that influence their predictive accuracy in political contexts.