LLM as a Mastermind: A Survey of Strategic Reasoning with Large Language Models

arXiv:2404.01230v1 Announce Type: new
Abstract: This paper presents a comprehensive survey of the current status and opportunities for Large Language Models (LLMs) in strategic reasoning, a sophisticated form of reasoning that necessitates understanding and predicting adversary actions in multi-agent settings while adjusting strategies accordingly. Strategic reasoning is distinguished by its focus on the dynamic and uncertain nature of interactions among multi-agents, where comprehending the environment and anticipating the behavior of others is crucial. We explore the scopes, applications, methodologies, and …

abstract adjusting agent arxiv cs.cl current form language language models large language large language models llm llms mastermind multi-agent opportunities paper reasoning strategies survey type understanding