西南石油大学学报(社会科学版) ›› 2026, Vol. 28 ›› Issue (5): 65-76.

• 人工智能与社会发展 • 上一篇    下一篇

思政大模型知识生产:图景呈现、现实隐忧和机制构建

米华全, 罗歆悦   

  1. 电子科技大学马克思主义学院, 四川 成都 611731
  • 发布日期:2026-10-29
  • 作者简介:米华全,电子科技大学马克思主义学院副教授,博士,i思政大模型创新实验室副主任,习近平文化思想研究中心四川省协同研究基地研究员,研究方向:思想政治教育。
  • 基金资助:
    四川省哲学社会科学基金马克思主义理论研究和建设工程项目“人工智能大模型赋能高校思政课教学模式改革与场景构建研究”(SCJJ25MGC11)。

Knowledge Production by Large Models for Ideological and Political Education: Landscape Presentation, Practical Concerns, and Mechanism Construction

MI Huaquan, LUO Xinyue   

  1. School of Marxism, University of Electronic Science and Technology of China, Chengdu Sichuan, 611731, China
  • Published:2026-10-29

摘要: 知识生产是思想政治教育内容创新的重要途径,人工智能大模型技术的快速发展引发了知识生产模式的深刻变革。思想政治教育垂类大模型依托数据驱动构建全景式数字资源池与动态化知识图谱,凭借智能算法实现知识挖掘、生成与呈现的全链条优化,通过交互学习打造人机协同的知识共创生态,为思政教育知识生产提供了技术赋能与创新支撑。然而,技术赋能亦伴随结构性风险:数据噪声与“算法黑箱”引发内容失真与质量失衡,语料偏见与算法局限导致价值失范与主流意识形态偏移,智能替代与技术异化消解主体能动性,数据流转与信息重组则诱发隐私侵犯与知识产权归属模糊等合规性问题。为了防范上述风险,需要从语料数据治理优化机制、主流意识形态引领机制、主体智能素养提升机制以及数据安全体系建设机制四个维度协同发力,构建系统化的应对机制,不断优化思政大模型知识生产的智能生态,助力思想政治教育知识生产的创新跃迁与全面提升。

关键词: 人工智能, 思政大模型, 思想政治教育, 知识生产, 数字思政, 精准思政, 智能思政

Abstract: Knowledge production is a critical avenue for content innovation in ideological and political education. The rapid advancement of artificial intelligence large model technology has triggered profound transformations in knowledge production models. Vertical large models for ideological and political education, driven by data, construct panoramic digital resource pools and dynamic knowledge graphs. Leveraging intelligent algorithms, they optimize the entire chain of knowledge mining, generation, and presentation, while fostering a human-machine collaborative knowledge co-creation ecosystem through interactive learning, thereby providing technological empowerment and innovative support for knowledge production in ideological and political education. However, technological empowerment is also accompanied by structural risks: data noise and "algorithmic black boxes" lead to content distortion and quality imbalance; corpus bias and algorithmic limitations give rise to value anomie and deviation from mainstream ideology; intelligent substitution and technological alienation undermine human agency; and data flow and information recombination trigger compliance issues such as privacy infringement and ambiguous attribution of intellectual property rights. To prevent and mitigate these risks, it is necessary to synergistically develop a systematic response mechanism across four dimensions: optimizing the corpus data governance mechanism, strengthening the mainstream ideology guidance mechanism, enhancing the mechanism for improving the intelligent literacy of education subjects, and constructing a data security mechanism. This will continuously optimize the intelligent ecosystem of knowledge production by large models for ideological and political education, facilitating innovative leaps and comprehensive improvements in ideological and political education.

Key words: artificial intelligence, large models for ideological and political education, ideological and political education, knowledge production, digital ideological and political education, precision ideological and political education, intelligent ideological and political education

中图分类号: