Integrating Large Language Models into Agent Models for Multi-Agent Simulations: Preliminary Report

There have been active attempts to integrate agents backed by large language models into various intelligent systems. In this paper, we describe our work on integrating an LLM into an agent model for multi-agent simulations (MASs). One long-term goal in implementing an MAS has been constructing a computational model that accurately simulates fine-grained human behaviors within the target environment. Building a model capable of capturing and reproducing the individual characteristics of a diverse range of people has been challenging, both in terms of implementation cost and complexity. We propose a method that uses an LLM to generate behavior individuality, and enables on the spot decision making based on the surrounding environment. We implement an MAS that uses agents based on the proposed method and verify the validity of their behaviors.