人工智能(Artificial Intelligence,以下简称AI)与科学研究的深度融合正驱动科学知识生产方式发生系统性变革,深入理解这一机制对推动我国科技创新与发展新质生产力具有重要意义。本研究以中国科大“机器化学家”平台为载体开展案例研究,通过案例数据编码分析和理论研究,(1)提出了AI驱动科学知识生产方式变革的四维协同机制。AI嵌入科学知识生产实践后对认知模式、工具形态、组织模式与制度体系进行了整体性重构,而非在单一技术要素维度的叠加。(2)研究发现,在AI驱动的科学知识生产实践中,认知、技术、组织、制度四个单元的内涵和结构呈现新的特征。认知层面,推动科学推演由经验导向向模型驱动转变,拓展科研人员的可解问题空间与认知边界;技术层面,驱动科研工具从离散执行向集成化智能平台演进;组织层面,人机协同创新促进协作网络从学科封闭走向跨学科开放协同;制度层面,评价激励、资源配置与数据治理规则动态适配智能化科研需求。(3)AI驱动的科学知识生产实践中,认知、技术、组织、制度之间互动关系呈现出新的内涵与形态。这种互动关系并非自然形成,而是在应对AI带来的一系列张力和冲突中不断调适,最终推动科学知识生产迈向“人—机—知识”的协同系统。研究拓展了科学知识生产方式及其系统结构理论在智能时代的适用边界与解释力,为加速以人工智能引领科研范式变革提供重要支撑。

The deep integration of artificial intelligence (AI) into scientific research is driving a systemic transformation in the mode of scientific knowledge production. Understanding this mechanism is essential for advancing technological innovation and developing new quality productive forces in China. This study conducts a case study of the Robotic AI-Chemist platform at the University of Science and Technology of China. Drawing on case study and theoretical analysis, the study makes three contributions. First, it proposes a four-dimensional synergistic mechanism through which AI transforms the mechanism of scientific knowledge production. Once embedded in the practice of scientific knowledge production, AI holistically reconfigures researchers' cognitive patterns, technology-tool forms, organizational structures, and institutional systems, rather than merely adding to a single technological dimension. Second, the study finds that the four units of knowledge production—cognition, technology, organization, and institution—exhibit new features in both content and structure. At the cognitive level, AI shifts scientific reasoning from experience-oriented to algorithm-driven, and expands researchers' solvable problem space and cognitive boundaries. At the technological level, AI drives the evolution of research tools from discrete execution toward integrated intelligent platforms. At the organizational level, human-machine collaborative innovation moves collaboration networks from disciplinary closure toward interdisciplinary open collaboration. At the institutional level, rules for evaluation and incentives, resource allocation, and data governance dynamically adapt to the demands of intelligent research. Third, the interactions among cognition, technology, organization, and institution take on new content and forms. These interactions do not emerge naturally. Instead, they are continuously adjusted in response to a series of tensions and conflicts brought by AI, ultimately advancing scientific knowledge production toward a human-machine-knowledge synergistic system. This study extends the applicability and explanatory power of theories on the mode of scientific knowledge production and its system structure in the age of intelligence. It provides important support for accelerating the AI-led transformation of research paradigms.