中国科技期刊研究 ›› 2025, Vol. 36 ›› Issue (7): 824-834. doi: 10.11946/cjstp.202504210399

• AI赋能学术期刊出版专题 • 上一篇    下一篇

中西方期刊界对生成式人工智能参与科研论文写作的风险治理比较分析

李金正1,2)()(), 李德1)   

  1. 1) 重庆大学新闻学院,重庆市高新区大学城南路55号 401331
    2) 重庆大学出版科学研究所,重庆市沙坪坝区大学城西路21号 401331
  • 收稿日期:2025-04-21 修回日期:2025-05-23 出版日期:2025-07-31 发布日期:2025-07-31
  • 作者简介:

    李金正(ORCID:0009-0009-2127-3159),博士,副教授,研究员,博士研究生导师,E-mail:;

    李 德,博士研究生。

    作者贡献声明: 李金正:提出论文选题、研究框架和研究方法指导,论文主要撰稿人; 李 德:样本数据收集与整理,参与论文初稿撰写和文献格式整理。
  • 基金资助:
    国家社科基金项目“生成式人工智能应用于学术成果写作出版的标准规范与治理研究”(24BTQ042)

Comparative analysis of risk governance in generative AI participation in academic paper writing between Chinese and Western journals

LI Jinzheng1,2)()(), LI De1)   

  1. 1) School of Journalism, Chongqing University, 55 South Daxuecheng Road, High-Tech District, Chongqing 401331, China
    2) Institute of Publishing Science, Chongqing University, 21 West Daxuecheng Road, Shapingba District, Chongqing 401331, China
  • Received:2025-04-21 Revised:2025-05-23 Online:2025-07-31 Published:2025-07-31

摘要:

【目的】系统比较中西方期刊界对GenAI参与科研论文写作的风险治理,为AI冲击下的国内学术伦理风险治理提供启示和建议。【方法】通过扎根理论对西方300家权威期刊的官方声明进行三级编码,进而就其核心范畴与中国期刊的相应文本展开平行比较。【结果】中西方期刊界均已意识到GenAI参与科研论文写作存在多方面的伦理和法律风险,并试图通过主体归责和伦理规约2条路径来实现风险治理,但在具体内涵和规范上存在一定的偏差;国内期刊存在着官方声明发布量偏低、西方公认的可使用边界未能明示、缺乏统一的声明披露格式、对于提交第三方检测报告以及GenAI使用程度的限定缺乏合理性说明等有待提高或完善的问题。【结论】根据“借鉴西方先进经验与扎根中国现实相结合”的原则,西方期刊界在GenAI参与科研论文写作的知识产权和隐私权保护意识、因算法偏见引发的社会舆情预警、对编辑主体的归责认定、对学术期刊社会义务的担当,以及对于模糊边界的解释条款等方面值得国内同行权衡借鉴。

关键词: 生成式人工智能, 科研论文, 风险治理, 中西比较

Abstract:

[Purposes] This study aims to systematically compare how Chinese and Western academic journals govern the risks associated with generative AI (GenAI) participation in academic paper writing, providing insights and recommendations for domestic academic ethics under the impact of AI. [Methods] Using grounded theory, the authors conducted a three-level coding of official statements from 300 leading Western journals and performed a parallel comparison of their core categories with corresponding texts from Chinese journals. [Findings] Both Chinese and Western academic journals have recognized the multifaceted ethical and legal risks of GenAI participation in academic writing and are attempting to manage these risks through two main approaches: subject accountability and ethical regulation. However, discrepancies exist in the specific content and norms. The study also identifies several issues in Chinese journals, such as a low volume of official statements, lack of clear disclosure of usage boundaries recognized in the West, absence of a unified declaration format, insufficient explanation of third-party detection report submission requirements, and unreasonable limitations on the extent of GenAI use. [Conclusions] Following the principle of “drawing on advanced Western experience while grounding in China’s reality”, the study suggests that Chinese academic journals should consider adopting Western practices in areas such as intellectual property and privacy protection in GenAI-assisted writing, early warning mechanisms for public opinion triggered by algorithmic bias, accountability assignment for editorial entities, fulfillment of journals’ social obligations, and interpretive clauses for ambiguous usage boundaries.

Key words: Generative artificial intelligence, Academic paper, Risk governance, Comparison between China and the West