生成式人工智能赋能科研知识生产研究述评

Research Review of Generative Artificial Intelligence Empowering Knowledge Production in Scientific Research

梳理国内外生成式人工智能赋能科研知识生产的研究成果,以期为其理论和实践应用提供参考。本研究主要采用文献计量与内容分析方法,并借助VOSviewer软件从知识生产与传播、生成式人工智能赋能、社会影响与变革、治理路径等四个方面综述国内外生成式人工智能赋能科研知识生产的研究现状。目前生成式人工智能赋能科研知识生产的研究取得了明显进展,但未来还需要深入探索并加强生成式人工智能对科研知识生产的影响因素、赋能科研知识生产全流程、科研知识生产的治理路径,以及拓展与丰富生成式人工智能赋能科研知识生产研究方法等方面的研究。

In the era of artificial intelligence,large language models represented by Generative Pre-trained Transformer(GPT)is penetrating in scientific research,motivating innovations in various sectors.In order to explore the research hotspots and future directions of generative AI-enabled scientific research knowledge production and bridge the gaps in current review research on this topic,this paper conducted a systematic review on the theoretical and practical application of generative AI(GenAI)in scientific research knowledge production based on 103 research papers published in China and 87 in other countries obtained from CNKI and Web of Science(WoS).With the methods of bibliometrics and keywords cluster analysis,it mapped the research trends with the VOSviewer by co-word analysis based on keywords,statistically outlined the relationship of each research topic and summarized the research hotspots by further literature review.Study revealed several research themes including the theoretical and practical application of GenAI in scientific research knowledge production,the produce and disseminate of knowledge,the impact of GenAI on scientific research process and the regulation of GenAI-enabled innovation.By comparison,researches in China emphasized more on the importance of AI ethics as well as proper regulations to scrutinize GenAI.Research review of the paper identifies the focus of recent research including:(1)The impact of GenAI on knowledge production and dissemination,human-machine collaboration and interaction.(2)GenAI as a supportive tool to innovate scientific research process such as to assist literature review,to help with research design and experimental plan,to conduct data analysis and to facilitate scientific writing and accelerate academic publishing.(3)The transformative changes in scientific paradigms and innovation in other sectors led by GenAI along with the concerns of academic misconduct,research integrity and other risks addressing AI ethics.(4)The call for proper regulations such as to enact GenAI scholarship standards from the perspective of law-makers,press sectors and technological methods,to improve regulatory accountability mechanism towards GenAI-generated contents,to upgrade GTP with high-quality and open-source techniques to minimize the possibilities of biased analysis,plagiarism and copyright infringement.In future outlook,this paper proposes further studies in the following spheres include to conduct research of GenAI according to the scientific research life circle covering the full process of knowledge production,to analyze the impact factors and underlying risks of GenAI on scientific research in different contexts,to broaden the research methods of GenAI with more case studies and to develop a proper regulatory framework towards GenAI to regulate the knowledge production and scientific research innovation in the context of AI.