生成式AI辅助情报学研究的场景、生成式行动者和挑战OACHSSCD

[Purpose/Significance]The rapid development of Generative Artificial Intelligence(GenAI)presents pro-found opportunities and complex challenges for the field of information science.To address this transformation effectively,it is essential to systematically investigate the methods,advantages,and limitations of GenAI-assisted information science research.Responding to the critical need for a structured framework to guide researchers in this new environment,this paper aims to provide both a theoretical foundation and practical guidance for the intelligent evolution of the discipline,and to empower scholars to harness the full potential of GenAI to enhance research efficiency and foster academic innovation while navigating its inherent complexities and risks.[Method/Process]This paper adopted a research design that combined theoretical construction with empirical validation.First,drawing on academic literature from the AI for Social Science(AI4SS)paradigm and integrating it with practical research experience,the paper systematically identified and summa-rized the application modes and scenarios of GenAI-assisted information science research and provided representative examples.Subsequently,the paper conducted two empirical studies.In the first study,which focused on data-centric AI-driven research,the paper developed an agentic workflow to intelligently identify the national affiliations of authors'institu-tions for datasets on the Dryad data publishing platform.In the second study,the paper explored the use of generative agents by conducting an experiment on AIGC(AI-Generated Content)information avoidance behavior,and compared the analytical results of this silicon sample experiment with the conclusions of existing studies that relied on traditional human sampling methods.Finally,the paper discussed the challenges and corresponding countermeasures for information science research under the AI4SS paradigm.[Result/Conclusion]The paper identifies five primary application modes of GenAI in assisting information science research:1)interacting with GenAI directly via dialogue interfaces;2)embedding GenAI functions into traditional research tools;3)embedding traditional research tools within GenAI-integrated systems;4)orchestrating intelligent system components;and 5)forming an autonomous research loop.These applications span the entire research lifecycle,including literature review,research design,data collection and analysis,and manuscript writing and revision.The paper confirms that GenAI enables the development of agentic workflows that assist in data collection,processing,and analysis.These workflows can retrieve and batch-collect the metadata of datasets and their corresponding papers;complete,align,cross-validate,and correct metadata;and draft experimental analysis reports in natural lan-guage.Crucially,the analytical conclusions of the silicon-sample experiment align with those of traditional sampling experiments.This finding provides robust empirical support for the emerging AI synthetic data-driven research paradigm in information science and highlights its advantages in efficiency and reproducibility.