ChatPCG: Large Language Model-Driven Reward Design for Procedural Content Generation

TL;DR

ChatPCG introduces a large language model-based framework that automates reward design for game AI, leveraging human insights to improve content generation and streamline development in multiplayer gaming.

Contribution

It presents a novel LLM-driven reward design method that integrates game expertise, automating reward creation and enhancing content generation for multiplayer games.

Findings

LLM can understand game mechanics and content tasks

Automated rewards improve game content diversity

Streamlines game AI development process

Abstract

Driven by the rapid growth of machine learning, recent advances in game artificial intelligence (AI) have significantly impacted productivity across various gaming genres. Reward design plays a pivotal role in training game AI models, wherein researchers implement concepts of specific reward functions. However, despite the presence of AI, the reward design process predominantly remains in the domain of human experts, as it is heavily reliant on their creativity and engineering skills. Therefore, this paper proposes ChatPCG, a large language model (LLM)-driven reward design framework.It leverages human-level insights, coupled with game expertise, to generate rewards tailored to specific game features automatically. Moreover, ChatPCG is integrated with deep reinforcement learning, demonstrating its potential for multiplayer game content generation tasks. The results suggest that the proposed LLM exhibits the capability to comprehend game mechanics and content generation tasks, enabling tailored content generation for a specified game. This study not only highlights the potential for improving accessibility in content generation but also aims to streamline the game AI development process.