摘要
大语言模型正在引发一场知识生产方式的变革,深刻影响着人类认识世界和改造世界的路径。本文旨在探究大语言模型驱动的知识生产方式变革如何重构决策范式,构建一个揭示两者内在关联的整合性分析框架,为驾驭智能革命、推进人机协同的管理决策提供理论指引。研究系统梳理了知识生产与决策范式的共生演进脉络,深入剖析了大语言模型引致知识生产方式革命的内在机理及由此带来的决策范式变革路径。研究发现:大语言模型通过知识表征、映射、生成、建模等技术环节,推动从离散符号到连续向量、从局部特征到全局表征的知识生产范式转变,催生了规模化的知识整合与交互式的知识调用能力。在重塑决策范式方面,这一知识生产革命通过重构知识供给方式、赋能人机混合智能、催生满意化进化决策等路径引领组织走向开放、复杂适应和动态演化。进一步地,研究为缓解“哈耶克难题”提供了技术思路。同时,研究也辩证分析了大语言模型应用中的技术风险与社会影响,最终提出了多层次的应用策略与实践建议,为人机协同、群体智慧涌现及决策范式变革拓展了新的可能性空间。
Large Language Models(LLMs)are initiating a revolutionary transformation in knowledge production methodologies,profoundly influencing how humans perceive and transform the world.This paper aims to explore how knowledge production transformations driven by LLMs reconstruct decision-making paradigms,establishing an integrative analytical framework that reveals their intrinsic connection and providing theoretical guidance for navigating the intelligent revolution and advancing human-machine collaborative management decisions.The research systematically examines the co-evolutionary trajectory of knowledge production and decision-making paradigms,and analyzes in depth the internal mechanisms through which LLMs revolutionize knowledge production methodologies and the consequent pathways of decision paradigm transformation,Findings reveal that LLMs,through technical processes including knowledge representation,mapping,generation,and mod eling,drive a paradigm shift from discrete symbols to continuous vectors and from local features to global representations,fostering large-scale knowledge integration and interactive knowledge retrieval capabilities.In terms of reshaping decision paradigms,this knowledge production revolution leads organizations toward openness,complex adaptation,and dynamic evolution through reconstructing knowledge supply mechanisms,enabling human-machine hybrid intelligence,and fostering satisficing evolutionary decision-making.Simultaneously,the research provides technical approaches to mitigating the"Hayek Problem"-the structural contradiction between distributed knowledge and centralized decisionmaking.The study also dialectically analyzes the technical risks and social impacts of LLM applications,ultimately proposing multi-level application strategies and practical recommendations that expand new possibility spaces for human-machine collaboration,collective wisdom emergence,and decision paradigm transformation.
作者
米加宁
李大宇
董昌其
Mi Jianing;Li Dayu;Dong Changqi(School of Economics and Management,Beijing University of Posts and Telecommunications;School of Finance and Public Administration,Harbin University of Commerce;School of Management,Harbin Institute of Technology)
出处
《管理世界》
北大核心
2025年第7期40-58,72,共20页
Journal of Management World
基金
国家社会科学基金重大项目“数据科学对社会科学转型的重大影响研究”(基金号:17ZDA030)的资助。
关键词
大语言模型(LLM)
知识生产方式
哈耶克难题
人机协同决策
决策范式
large language models(LLMs)
knowledge production methods
Hayek Problem
human-machine cllaborative decision-making
decision-makingparadigm
