The paper "AI for Social Science and Social Science of AI: A Survey" by Ruoxi Xu et al. explores the intersection of artificial intelligence (AI) and social science, categorizing previous research into two main directions: AI for social science and social science of AI. The authors provide a comprehensive review of the current state of these fields, highlighting their technical approaches, research objectives, and applications. In the first direction, AI for social science, large language models (LLMs) are used to enhance various stages of social science research, such as literature review, hypothesis generation, data analysis, and writing assistance. The paper discusses the potential of LLMs in improving efficiency and providing valuable insights, while also addressing limitations and ethical concerns. The second direction, social science of AI, examines AI agents as social entities with human-like cognitive and linguistic capabilities. This involves studying the behavioral laws of AI agents from a social science perspective, particularly in sub-disciplines like psychology, sociology, economics, politics, and linguistics. The paper also compiles available simulation tools that use LLMs to simulate diverse social situations and behaviors. The authors conclude by emphasizing the importance of understanding the distinctions and connections between these two directions, providing a cohesive framework for researchers to navigate the field. They also highlight the potential future directions, including integrating domain knowledge, developing high-reward prompt strategies, and expanding context windows for LLMs. The paper aims to facilitate further research and applications in both areas, contributing to a deeper understanding of the relationship between AI and social science.The paper "AI for Social Science and Social Science of AI: A Survey" by Ruoxi Xu et al. explores the intersection of artificial intelligence (AI) and social science, categorizing previous research into two main directions: AI for social science and social science of AI. The authors provide a comprehensive review of the current state of these fields, highlighting their technical approaches, research objectives, and applications. In the first direction, AI for social science, large language models (LLMs) are used to enhance various stages of social science research, such as literature review, hypothesis generation, data analysis, and writing assistance. The paper discusses the potential of LLMs in improving efficiency and providing valuable insights, while also addressing limitations and ethical concerns. The second direction, social science of AI, examines AI agents as social entities with human-like cognitive and linguistic capabilities. This involves studying the behavioral laws of AI agents from a social science perspective, particularly in sub-disciplines like psychology, sociology, economics, politics, and linguistics. The paper also compiles available simulation tools that use LLMs to simulate diverse social situations and behaviors. The authors conclude by emphasizing the importance of understanding the distinctions and connections between these two directions, providing a cohesive framework for researchers to navigate the field. They also highlight the potential future directions, including integrating domain knowledge, developing high-reward prompt strategies, and expanding context windows for LLMs. The paper aims to facilitate further research and applications in both areas, contributing to a deeper understanding of the relationship between AI and social science.