Scholarlink

<p>The recent surge in artificial intelligence, exemplified by large language models (LLMs), has created both opportunities and challenges for methodological innovation in political science. This has sparked discussions about a potential paradigm shift in the social sciences. As such, how might social scientists comprehensively understand the influence of LLMs on the production and transformation of knowledge in this field from an integrated perspective that encompasses both technology and methodology? What are the specific applications of LLMs in the development of innovative methods in political science research? This paper proposes the Intelligent Computing Social Modeling (ICSM) framework as a systematic approach to addressing these issues through a comprehensive analysis of LLM mechanisms. ICSM capitalizes on the capabilities of LLMs in idea synthesis and action simulation, thereby facilitating intellectual exploration in political science through “simulated society construction” and “simulation validation.” We empirically demonstrate the operational pathways and methodological advantages of ICSM through a simulation of the U.S. presidential election. By integrating traditional social science paradigms, ICSM enhances the quantitative paradigm’s ability to leverage big data for assessing variable impacts while also providing the qualitative paradigm with empirical evidence for mechanism discovery at the individual level. This methodology achieves a balance between interpretability and predictive accuracy in social science research. Our framework offers practical tools for analyzing complex social phenomena using large language models.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Intelligent Computing Social Modeling and Methodological Innovations in Political Science in the Era of Large Language Models

  • 期刊论文

  • 发表日期：

    2025年4月26日

查看全文
  • Zhenyu Wang,
  • Dequan Wang,
  • Yi Xu,
  • Lingfeng Zhou,
  • Yiqi Zhou

摘要

The recent surge in artificial intelligence, exemplified by large language models (LLMs), has created both opportunities and challenges for methodological innovation in political science. This has sparked discussions about a potential paradigm shift in the social sciences. As such, how might social scientists comprehensively understand the influence of LLMs on the production and transformation of knowledge in this field from an integrated perspective that encompasses both technology and methodology? What are the specific applications of LLMs in the development of innovative methods in political science research? This paper proposes the Intelligent Computing Social Modeling (ICSM) framework as a systematic approach to addressing these issues through a comprehensive analysis of LLM mechanisms. ICSM capitalizes on the capabilities of LLMs in idea synthesis and action simulation, thereby facilitating intellectual exploration in political science through “simulated society construction” and “simulation validation.” We empirically demonstrate the operational pathways and methodological advantages of ICSM through a simulation of the U.S. presidential election. By integrating traditional social science paradigms, ICSM enhances the quantitative paradigm’s ability to leverage big data for assessing variable impacts while also providing the qualitative paradigm with empirical evidence for mechanism discovery at the individual level. This methodology achieves a balance between interpretability and predictive accuracy in social science research. Our framework offers practical tools for analyzing complex social phenomena using large language models.