Overcoming the“Collingridge Dilemma”-Ethical Issues of Scientific Data in the Era of Large Models:Types,Essence and the“Tao”&“Technique”for Solution
Hu Feng1, Zeng Guanxiu2, Han Bo1
1. Jiangsu Information Institute of Science and Technology(Jiangsu Academy of Science and Technology for Development),Nanjing 210042,China; 2. School of Political Science & Law, Shaoguan University,Shaoguan 512005,China
Abstract：In the era of large models marked by the ubiquitous application of AIGC,scientific data has become a core production factor driving innovation,but it has also given rise to serious ethical issues.To overcome the“Colingridge dilemma”of ethical governance and regulation,the paper uses the iceberg model as a theoretical tool to identify the explicit and implicit issues of scientific data ethics,and summarizes the essence of these issues.From a micro-level perspective,ethical issues stem from the deep alienation of the relationship between people,the relationship between people and data,and the relationship between people and society driven by technology.From a macro-level ecological perspective,ethical issues are the result of deep conflicts between technological logic,value logic,and governance logic.Based on this,a synergy of“Tao”and“Technique”solution is proposed.For the ethical issues of scientific data in the era of large models,the“Tao”focuses on four main action principles:advocating truth-seeking and pragmatism;and ensuring data fairness;upholding shared responsibility;enhancing public welfare.Improving institutional arrangements,strengthening technological empowerment,optimizing platform design,and deepening capacity building become the“Technique”for resolving ethical issues.
胡峰, 曾关秀, 韩博. 走出“科林格里奇困境”——大模型时代科学数据伦理问题:类型、本质及破解的“道”与“术”[J]. 中国科技论坛, 2026(7): 20-28.
Hu Feng, Zeng Guanxiu, Han Bo. Overcoming the“Collingridge Dilemma”-Ethical Issues of Scientific Data in the Era of Large Models:Types,Essence and the“Tao”&“Technique”for Solution. , 2026(7): 20-28.
[1] 西格蒙德·弗洛伊德.文明及其不满[M].林宏涛,译.杭州:浙江大学出版社,2024. [2] 张乐,赵爽爽.人工智能的科林格里奇困境与安全监管的敏捷性调适[J].中国行政管理,2026,42(3):38-46. [3] 朱明婷,徐崇利.数据伦理全球治理初探:现状、问题与对策:以各类文件为样本分析[J].海南大学学报(人文社会科学版),2025,43(6):158-169. [4] 夏永红.人工智能伦理治理范式:从价值对齐到价值共生[J].自然辩证法通讯,2025,47(1):1-8. [5] 刘鑫怡,徐峰,司伟攀.生成式人工智能应用伦理风险的形成机理及治理策略研究[J].中国软科学,2025(10):194-204. [6] 刘先瑞,司莉.科学数据伦理治理:政策框架与路径:以英国为例[J].现代情报,2025(1):112-123,134. [7] 李天硕,徐琳琳,张冬荣,等.基于政策文本分析的科研数据伦理治理框架构建[J].中国科技期刊研究,2025,36(1):1-17. [8] 肖巍.人体生物数据伦理风险治理的中国智慧[J].求索,2025(4):115-122,207. [9] 李宗富,姜爱玲.档案数据治理的伦理审视:内蕴、风险与路径[J].档案学研究,2024(4):4-11. [10] 梁宇,郑易平.科研数据共享中的数据伦理问题研究[J].中国科技论坛,2022(1):22-27. [11] Floridi L,Taddeo M.What is data ethics[J].Philosophical Transactions of The Royal Society of A:Mathematical,Physical and Engineering Sciences,2016,374(2083):20160360. [12] Hasselbalch G.Making sense of data ethics:the powers behind the data ethics debate in European policymaking[J].Internet Policy Review,2019,8(2):1-19. [13] 冯薇,石庆功,李一然.英国政府数据伦理框架的制定、实施及启示[J].图书馆学研究,2024(3):43-49. [14] 赵一鸣,谭欣佩,张胜发.科学数据价值评估研究[J].图书情报工作,2025,69(6):58-71. [15] 张楚辉,裴雷,李卓卓.我国公共数据开放中的数据伦理:生成逻辑、问题识别与敏捷治理[J].情报学报,2026,45(2):180-192. [16] Thordal-Christensen H.A holistic view on plant effector-triggered immunity presented as an iceberg model[J].Cellular and Molecular Life Sciences,2020,77(20):3963-3976. [17] 赵伟,任潇宁,薛吟兴.基于集成学习的成员推理攻击方法[J].信息网络安全,2024,24(8):1252-1264. [18] Li Na,Zhou Chunyi,Gao Yansong,et al.Machine unlearning:taxonomy,metrics,applications,challenges,and prospects[J].IEEE Transactions on Neural Networks and Learning Systems,2025,36(8):13709-13729. [19] Pasquetto I V,Cullen Z,Thomer A,et al.What is research data“misuse”?And how can it be prevented or mitigated[J].Journal of the Association for Information Science and Technology,2024,75(12):1413-1429. [20] 罗敬蔚,李勇坚.人工智能大语言模型幻觉风险与制度因应[J].科学管理研究,2026,44(1):80-90. [21] Alkaissi H,Mcfarlane S I.Artificial hallucinations in Chat-GPT:implications in scientific writing[J].Cureus,2023,15(2):e35179. [22] 金耀.人工智能模型训练的合成数据治理路径[J]法学论坛,2026,41(2):115-126. [23] 徐伟,韦红梅.生成式人工智能训练数据风险治理:欧盟经验及其启示[J]现代情报,2025,45(5):89-98. [24] Zhang Chengjun,Ren Zhengju,Xiang Gaofeng,et al.A comprehensive comparative analysis of publication monopoly phenomenon in scientific journals[J].Journal of Informetrics,2025,19(1):101628. [25] 顾男飞.人工智能数据垄断风险预警及治理[J]情报杂志,2025,44(12):77-84. [26] 孙瑜晨,郑诗彦.开放共享为导向的科学数据分类分级制度建构:价值旨归与实现路径[J].科技进步与对策,2025,42(12):140-150. [27] 胡峰,黎亮,姚怡帆.“我的健康数据我做主”:欧洲健康数据空间运营的机理刻画、挑战分析及策略因应[J].图书与情报,2025(5):133-144. [28] 闫坤如.人工智能技术异化及其本质探源[J].上海师范大学学报(哲学社会科学版),2020,49(3):100-107. [29] 拉埃尔·耶吉.异化.田毅松,凤秀林,译.北京:北京师范大学出版社,2025. [30] 张恒力,李昂.科技伦理生态何以可能[J].自然辩证法研究,2025,41(7):3-8,53. [31] 哈里·柯林斯,特雷弗·平奇.勾勒姆医生:如何理解医学[M].雷瑞鹏,译.上海:上海人民出版社,2022. [32] 金惠敏.由“术”而“道”:老庄整体性技术观研究[J].哲学研究,2015(6):55-62. [33] 董慧,杜晓依.新时代超大城市治理现代化的“道”与“术”[J].甘肃社会科学,2024(3):116-123. [34] 霍朝光,赵栋祥.数据公平:数字经济时代社会新命题[J].情报学报,2026,45(2):165-179. [35] 胡峰.流动的丰盈:面向可信数据空间建设的理论观照与实践观察[J].图书情报知识,2025,42(5):6-18,65.
2026, 
