陈凯华,李赫扬,刘泓欣,赵彬彬,杨硕.人工智能赋能科学研究：内涵、特征与体系[J].中国科学院院刊,2026,41(6):1206-1219.

人工智能赋能科学研究：内涵、特征与体系

Artificial intelligence for science: Connotations, characteristics, and system

陈凯华^1,2
中国科学院大学 公共政策与管理学院 北京 100049;中国科学院大学 国家前沿科技融合创新研究中心 北京 100049
CHEN Kaihua^1,2
School of Public Policy and Management, University of Chinese Academy of Sciences, Beijing 100049, China;Research Center for National Frontier S&T Integration and Innovation, University of Chinese Academy of Sciences, Beijing 100049, China
李赫扬^1,2
中国科学院大学 公共政策与管理学院 北京 100049;中国科学院大学 国家前沿科技融合创新研究中心 北京 100049
LI Heyang^1,2
School of Public Policy and Management, University of Chinese Academy of Sciences, Beijing 100049, China;Research Center for National Frontier S&T Integration and Innovation, University of Chinese Academy of Sciences, Beijing 100049, China
刘泓欣^1,2
中国科学院大学 公共政策与管理学院 北京 100049;中国科学院大学 国家前沿科技融合创新研究中心 北京 100049
LIU Hongxin^1,2
School of Public Policy and Management, University of Chinese Academy of Sciences, Beijing 100049, China;Research Center for National Frontier S&T Integration and Innovation, University of Chinese Academy of Sciences, Beijing 100049, China
赵彬彬²
中国科学院大学 国家前沿科技融合创新研究中心 北京 100049
ZHAO Binbin²
Research Center for National Frontier S&T Integration and Innovation, University of Chinese Academy of Sciences, Beijing 100049, China
杨硕^1,2*
中国科学院大学 公共政策与管理学院 北京 100049;中国科学院大学 国家前沿科技融合创新研究中心 北京 100049
YANG Shuo^1,2*
School of Public Policy and Management, University of Chinese Academy of Sciences, Beijing 100049, China;Research Center for National Frontier S&T Integration and Innovation, University of Chinese Academy of Sciences, Beijing 100049, China

         人工智能赋能科学研究;科研范式;人机协同;支撑体系

        artificial intelligence for science (AI4S);scientific research paradigm;human-machine collaboration;supporting framework

        人工智能正在深刻改变科学研究的基本内涵，重塑其知识生产方式和组织运行方式，推动形成人工智能赋能科学研究的新型科研范式，并加速全链条创新范式变革。文章从赋能应用、工具方法和认识知识3个维度，界定人工智能赋能科学研究的基本内涵，提炼其人机共生、自主演化、交叉融合、资源密集与开放生态5个核心特征，并构建涵盖基础设施、数据资源、模型工具、任务执行和应用场景等层次的支撑体系与架构。在此基础上，结合我国现实情况与战略需求，进一步提出夯实智能化科研基础设施底座、构建高质量科学数据供给体系、打造可信智能科研工具与开源协作生态、建立人机协同的智能科研执行体系、强化重大场景牵引与敏捷安全治理、推进适应智能化的科研组织模式变革等策略，为人工智能赋能科学研究实践提供理论与政策参考。

        Artificial intelligence is profoundly transforming the fundamental nature of scientific research, reshaping its modes of knowledge production and organizational operation, driving the emergence of a new artificial intelligence for science (AI4S) research paradigm, and accelerating full-chain innovation paradigm transformation. This study defines the basic connotations of AI4S across three dimensions, namely, enabling applications, tools and methods, and epistemic knowledge, and systematically identifies five core characteristics: human-machine symbiosis, autonomous evolution, interdisciplinary integration, resource intensity, and open ecosystems. It further constructs a supporting system and operational architecture encompassing layers of infrastructure, data resources, model tools, task execution, and application scenarios. Building on this foundation, and in light of China’s practical circumstances and strategic needs, the study proposes a set of policy recommendations: consolidating the intelligent scientific research foundation, building a high-quality scientific data supply system, developing trusted intelligent research tools and an open-source collaboration ecosystem, establishing human-machine collaborative execution mechanisms, strengthening agile and secure governance, and advancing the transformation of scientific research organizational models, to provide theoretical and policy reference for the practice of AI4S.