ChatGPT聚焦临床医学教育管理研究热点的可行性与伦理考量  

Feasibility and ethical considerations of ChatGPT focusing on research hotspots in clinical medical education management

作　　者：曾嘉霖 苏萍 黄昉菀[3] ZENG Jialin;SU Ping;HUANG Fangwan(Department of Clinical Medicine,Fujian Medical University,Fuzhou 350108,China;Fujian Medical University Union Clinical College,Fuzhou 350001,China;College of Computer and Data Science,Fuzhou University,Fuzhou 350108,China)

机构地区：[1]福建医科大学临床医学部,福建福州350108 [2]福建医科大学协和临床医学院,福建福州350001 [3]福州大学计算机与大数据学院,福建福州350108

出　　处：《中国医学伦理学》2024年第9期1108-1118,共11页Chinese Medical Ethics

基　　金：国家自然科学基金区域联合重点项目“跨网络跨媒体两岸热点事件的智能挖掘与推荐”(U21A20472);福建省本科高校教育教学研究项目(重大项目)“‘校-院-院’临床教学共同体‘多元一体互融’的构建与实践”(FBJY20230235)。

摘　　要：目的探讨利用大型语言模型ChatGPT聚焦临床医学教育管理领域中研究热点的可行性,旨在加速该领域科学研究的进程。方法首先选定该领域的六个关键课题,通过提问引导ChatGPT自动生成每个课题中最紧迫或最重要的五个研究热点;然后组织六名临床医学教育管理人员利用五分李克特量表对ChatGPT生成的研究热点从针对性、人文性、辩证性、拓展性和独创性五个维度进行评估;最后根据描述性统计、评分相似性、指标相关性对评估结果进行多角度分析。结果ChatGPT生成的研究热点在课题的针对性方面表现优异,在人文性、辩证性和拓展性方面的表现也令人满意,但在独创性方面则表现一般。结论ChatGPT可以作为聚焦临床医学教育管理中研究热点的辅助工具,但仍需更多努力来增强其所生成的研究热点的独创性。而在运用ChatGPT聚焦研究热点时,存在着一系列伦理风险,包括数据的虚假与滥用、算法的歧视与偏见、学术的失信与失范。研究者从技术层面可整合多种大型语言模型的聚焦结果,利用数据和算法多样性,规避伦理风险;从应用层面可利用个体鉴别和群体论证等手段,以批判性思维谨慎审视,规避伦理风险。Objective:To explore the feasibility of using the large-scale language model ChatGPT to focus on research hotspots in the field of clinical medical education management,aiming to accelerate the scientific research process in this field.Methods:First,six key topics in the field were selected,and ChatGPT was guided to automatically generate the five most urgent or important research hotspots in each topic through questioning.Then,six clinical medical education managers were organized to use the five-point Likert scale to evaluate the research hotspots generated by ChatGPT from five dimensions,including pertinence,humanity,dialectics,expansion,and originality.Finally,the evaluation results were analyzed from multiple perspectives based on descriptive statistics,score similarity,and indicator correlation.Results:The research hotspots generated by ChatGPT performed excellently in terms of topic specificity,and were also satisfactory in terms of humanistic,dialectical,and expansive aspects,but performed mediocrely in terms of originality.Conclusion:ChatGPT can serve as an auxiliary tool to focus on research hotspots in clinical medical education management,but more efforts are still needed to enhance the originality of the research hotspots it generates.However,when using ChatGPT to focus on research hotspots,there are a series of ethical risks,including false and abusive data,discrimination and bias in algorithms,as well as academic dishonesty and misconduct.From the technical perspective,researchers can integrate the focused results of multiple large-scale language models,and utilize data and algorithm diversity to avoid ethical risks.From the application perspective,individual identification,group argumentation,and other means can be utilized,and carefully examined with critical thinking to avoid ethical risks.

关 键 词：临床医学教育管理 研究热点 大型语言模型 伦理风险 规避进路 

分 类 号：R-052[医药卫生]