|
| 生成式人工智能使用披露政策的“清单”管控与“说明书”指引模式比较与启示 |
| Comparison and enlightenment of the"checklist"control model and the"instruction manual"guidance model for generative artificial intelligence usage disclosure policies |
| 投稿时间：2026-03-12 修订日期：2026-08-12 |
| DOI： |
| 中文关键词: 生成式人工智能 学术出版 政策 科研诚信 治理 |
| 英文关键词: Generative artificial intelligence Academic publishing Policy Research integrity Governance |
| 基金项目:河南省教科规划重大课题“河南省智慧教育体系构建与实践路径研究”(2026JKZB06),国家自然科学基金青年项目“AI-agent成员参与创业团队决策的全过程机理研究：问题导向搜索视角（72402060）” |
|
| 摘要点击次数: 157 |
| 全文下载次数: 0 |
| 中文摘要: |
| 【目的】生成式人工智能（GenAI）深度介入科研，全球学术出版界加速构建使用披露规范，但国内外政策范式存在显著差异。本文旨在提炼并比较“清单”管控与“说明书”指引两种典型模式，为我国科技期刊治理体系优化提供参考。【方法】选取19家中国科技期刊及5家国际出版集团的24份政策文本，采用政策文本分析法，从政策工具类型、作用阶段、政策力度三个维度进行系统编码与比较。【结果】两种模式呈现显著分野：“清单”模式条款简洁、立场权威，重在划定底线，守底线效率高但应对技术迭代灵活性不足；“说明书”模式内容详细、场景细分，强调操作指引与责任归属，但执行成本较高。【结论】建议我国科技期刊在坚守伦理底线基础上，借鉴场景化治理思路，推动政策向分层分类指引演进——保留核心禁令，建立分环节披露清单，明确各方责任，实现治理精准化与现代化，为全球科技期刊伦理建设提供中国方案。 |
| 英文摘要: |
| [Purposes]With generative artificial intelligence (GenAI) deeply embedded in scientific research, academic publishing worldwide is accelerating the development of disclosure norms. However, significant paradigmatic differences exist between domestic and international policies. This study aims to identify and compare two typical models:the "checklist" control model and the "instruction manual"guidance model, to inform the optimization of China’s scientific journal governance system. [Methods] A total of 24 policy documents from 19 Chinese scientific journals and 5 international publishing groups were selected. Using policy text analysis, systematic coding and comparison were conducted from three dimensions: policy tool type, stage of action, and policy intensity.[Findings] The two models exhibit distinct characteristics: the "checklist" model features concise clauses and an authoritative stance, offering high efficiency in setting bottom lines but lacking flexibility in responding to technological iteration; the "instruction manual" model provides detailed content and scenario-specific guidance, with clear operational instructions and accountability, though at higher implementation costs. [Conclusions] It is recommended that Chinese scientific journals, while adhering to ethical bottom lines, adopt a scenario-based governance approach and evolve toward a tiered and categorized guidance system. This entails retaining core prohibitions, establishing a stage-wise disclosure checklist, and clarifying the responsibilities of all parties, so as to achieve precise and modernized governance and contribute a Chinese model to the global ethical construction of scientific journals.
Keywords: Generative artificial intelligence; Academic publishing; Policy; Research integrity; Governance |
|
View Fulltext
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
