融合研究过程机制及实现路径基于美国融合加速器的多案例探索性分析

The Processing Mechanism and Implementing Path of Convergence Research:An Exploratory Multiple Case Analysis of the U.S.Convergence Accelerator

融合研究是一种科研新范式,各主要科技强国都在积极倡导和推动融合研究发展。美国融合加速器开创性地建立了与融合研究相适应的运行机制和管理模式。基于扎根理论的多案例比较研究方法,以美国融合加速器中3个与人工智能相关的融合研究轨道为研究对象,对融合研究过程机制进行探索性研究,并构建了融合加速器融合研究过程机制的理论模型。研究结果表明,融合研究存在两个核心过程机制:一是在融合研究各阶段中,形成了由融合动力、协作联动和利益联结所组成的融合研究自组织机制;二是在融合加速器目标约束和动态监督的调控与反馈中,形成了融合研究柔索驱动机制。研究结论为促进我国融合研究发展、应对人工智能时代所带来的发展机遇和治理挑战提供参考。

The National Science Foundation of the United States has established a Conver-gence Accelerator with the goal of developing a scientific management system that adapts to con-vergence research.To explore the governance model and process mechanisms of convergence re-search,this study employs the method of programmatic grounded theory and conducts a multi-case comparative research.It selects 70 projects from three research tracks related to artificial in-telligence within the Convergence Accelerator as research subjects and conducts an exploratory analysis of the process mechanisms and implementation paths of convergence research.Addition-ally,a theoretical model of the process mechanisms of convergence research within the Conver-gence Accelerator is constructed.The study reveals three main conclusions.Firstly,the model of process mechanisms for convergence research within the Accelerator consists of 13 primary cat-egories.Among these categories,"Converging Forces""Collaborative Linkages"and"Interest Conjunctions"constitute the core content of the self-organizing process of convergence research;"Objective Constraints"and"Dynamic Supervision"form the primary content of the management and service process for convergence research.Secondly,the Convergence Accelerator naturally develops a bottom-up self-organizing process mechanism for convergence research.It includes the core mechanism of"Utility-driven"convergence research,the"Collaboration-centered"con-vergence research mechanism,and the"Overflow-focused"convergence research benefit mecha-nism.Lastly,the study introduces the"Flexible-strand driving mechanism for convergence re-search",which aims at reveal the governance model formed by the interaction between objective constraints and dynamic supervision within the Convergence Accelerator.This mechanism oper-ates by leveraging top-down convergence research objectives as the guiding force and the dy-namic supervision of the Convergence Accelerator as the tension,leading to the formation of a stable governance state during the stages of convergence research I and II.The research findings provide valuable insights for promoting a paradigm shift in convergence research and addressing the opportunities and challenges of the AI era.First,it is crucial to promote the development of the convergence research community and facilitate the reciprocal interaction between knowledge communities and problem communities.Second,the establishment of a flexible-strand and bal-anced scientific management model is of paramount importance for the advancement of conver-gence research.In this model,the"rigidity"lies in the top-level design,specifically setting a clear vision for the construction of national strategic research forces.On the other hand,the"flex-ibility"is reflected in three aspects:"team-oriented"technology services,"convergence process-oriented"competitive management and"outcome-oriented"dynamic supervision.