LLM as a Mastermind: A Survey of Strategic Reasoning with Large Language Models
arxiv(2024)
摘要
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 evaluation metrics related to
strategic reasoning with LLMs, highlighting the burgeoning development in this
area and the interdisciplinary approaches enhancing their decision-making
performance. It aims to systematize and clarify the scattered literature on
this subject, providing a systematic review that underscores the importance of
strategic reasoning as a critical cognitive capability and offers insights into
future research directions and potential improvements.
更多查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要