科研管理

科研管理 ›› 2026, Vol. 47 ›› Issue (1): 1-10.DOI: 10.19571/j.cnki.1000-2995.2026.01.001

• 论文 •    下一篇

人工智能赋能的本质认识：数据、知识与系统的三重整合

潘教峰1,2，王楚扬2，吴静1,2   

  1. 1.中国科学院科技战略咨询研究院，北京100190；
    2.中国科学院大学公共政策与管理学院，北京100049
  • 收稿日期:2025-07-25 修回日期:2025-11-06 接受日期:2025-11-20 出版日期:2026-01-20 发布日期:2026-01-12
  • 通讯作者: 吴静
  • 基金资助:
    国家社会科学基金资助。

The essence of AI-enabled empowerment: A tripartite integration paradigm of data, knowledge and systems

Pan Jiaofeng1,2, Wang Chuyang2, Wu Jing1,2

  1. 1.Institutes of Science and Development, Chinese Academy of Sciences, Beijing 100190, China; 
    2. School of Public Policy and Management, University of Chinese Academy of Sciences, Beijing 100049, China
  • Received:2025-07-25 Revised:2025-11-06 Accepted:2025-11-20 Online:2026-01-20 Published:2026-01-12

PDF

55

摘要：    人工智能是典型的多学科、多领域融合的交叉学科，具有很强的渗透性与融合性。在“人工智能+”行动深入实施的背景下，从整合视角出发，深入剖析人工智能赋能的底层逻辑，系统厘清赋能本质机理，对明确人工智能与各领域深度融合的关键路径、推动人工智能应用落地、优化人工智能治理具有重要的理论和现实意义。本文提出：人工智能赋能的本质是“数据整合-知识整合-系统整合”的三重整合，即数据整合实现跨模态、跨时空、跨领域的数据融合，知识整合实现关联识别、因果推理、矛盾发现、收敛逼近、突变涌现五大能力的驱动，系统整合实现基础技术、功能技术和领域技术的工程化落地。针对整合中面临的数据流通机制尚不健全、算法偏见与决策偏差、系统安全脆弱性增大等问题，建议：完善数据流通制度和标准建设，强化全生命周期数据安全防护；加强算法偏差治理，提升模型透明度与可解释性；创新智能系统整合落地，提升工程韧性与伦理合规能力。本研究对促进人工智能的健康发展、推动产业智能化转型升级具有参考和借鉴价值。

关键词: 人工智能, 数据整合, 知识整合, 系统整合

Abstract:    Artificial intelligence is a typical interdisciplinary field that integrates multiple disciplines and domains, exhibiting strong pervasiveness and convergence. Against the backdrop of the deepening implementation of the "AI+" initiative, adopting an integrative perspective to dissect the underlying logic of AI-enabled empowerment and systematically clarifying its essential mechanisms is of paramount theoretical and practical significance. This effort is crucial for identifying key pathways for the deep integration of AI with various sectors, promoting the practical application of AI, and optimizing AI governance. Based on this, this paper proposed that the essence of AI-enabled empowerment lies in the triple integration of "data integration, knowledge integration, and system integration". Specifically: (1) Data integration enables the integration of cross-modality, cross-spatiotemporal, and cross-domain datasets; (2) Knowledge integration drives five core capabilities: association recognition, causal reasoning, contradiction discovery, convergence approximation mutation emergence; (3) System integration achieves the engineering implementation of foundational technologies, functional technologies, and domain-specific technologies. To address challenges in integration—including inadequate data circulation mechanisms, algorithmic bias and decision-making deviations, and increased system security vulnerabilities—the following recommendations are proposed: (1) data circulation frameworks and standardization protocols should be refined to strengthen full-lifecycle data security safeguards; (2) algorithmic bias governance should be enhanced to improve model transparency and interpretability; and (3) the implementation of intelligent system integration should be pioneered to advance engineering resilience and ethical compliance capabilities. This study will provide valuable insights and references for promoting the healthy development of artificial intelligence and driving the intelligent transformation and upgrading of industries.

Key words: artificial intelligence, data integration, knowledge integration, system integration