摘要
生成式人工智能(GAI)正在全球掀起新一轮技术革命,迅速成为推动经济增长、重塑劳动力市场和引发社会治理变革的重要力量。本文在梳理早期自动化到GAI的技术演进脉络基础上,系统评述了GAI作为通用技术与创新方法的双重技术属性突破,以及GAI对经济增长与劳动力市场的多维影响:GAI的技术突破使得AI的影响范围从物理领域延伸至认知领域,实现物质生产与知识创新的双重革新。理论层面,新古典增长理论与内生增长理论分别从任务自动化、新任务创造与知识生产自动化等路径,探讨了GAI带来持续增长或引发经济奇点的可能性;实证层面,微观生产率提升与宏观增长滞后之间的索洛悖论仍未完全破解。劳动力市场上,GAI通过替代、互补与创造三重机制重塑就业结构与收入分配,积极与负面效应并存。此外,GAI还可能引发隐私侵犯、市场垄断、信息伪造与伦理失序等潜在危害,同时也存在通过技术探索与制度设计迈向“图灵变换”的可能路径。最后,本文提出了一些未来研究需要深化探索的方向。
Generative artificial intelligence(GAI)is sparking a new global wave of technological revolution,rap⁃idly becoming a transformative force reshaping economic growth trajectories,labor market dynamics,and the social gov⁃ernance structure.As GAI quickly penetrates critical sectors such as education,healthcare,finance,and manufacturing,the systematic evaluation of its economic impact has emerged as a central concern for both academic researchers and poli⁃cymakers.This paper proposes a comprehensive analytical framework for understanding the multifaceted consequences of GAI,structured along three interrelated dimensions:technological evolution,economic effects,and institutional re⁃sponses.At the technological level,GAI systems exhibit unprecedented levels of autonomy,content-generation capability,and adaptability across a wide range of domains.Their functional scope has expanded beyond conventional physical and routine tasks into domains of cognition,language,and creativity.As a new generation of general-purpose technology(GPT),GAI fundamentally reconfigures value-creation processes across industrial,educational,and service landscapes.Notably,it demonstrates the capacity for“inventing the method of invention”-an innovation that redefines not just what is created,but how creation occurs.This dual characteristic positions GAI as both a productivity-enhancing tool and a methodology-shaping force,with profound implications for the future of knowledge production and economic transforma⁃tion.At the level of economic growth,GAI exerts complex and multidimensional impacts that span both theoretical and empirical realms.Theoretically,extensions of the neoclassical growth model by incorporating task automation and new task creation recognize GAI as a source of both capital-augmenting and labor-augmenting innovation,capable of overcom⁃ing the growth constraints traditionally associated with automation.The endogenous growth theory further emphasizes GAI’s role in the automation of knowledge production and R&D,which may give rise to exponential,“singularity-like”increases in productivity and innovation.In empirical terms,GAI faces a well-documented“productivity paradox”,namely,its benefits are visible at the micro level(e.g.,improved task performance and firm competitiveness)but remain largely undetectable in aggregate productivity data.This mismatch suggests that existing macroeconomic models and sta⁃tistical systems are ill-equipped to capture GAI’s latent effects,calling for more granular data infrastructure and updated theoretical paradigms capable of addressing technological complexity.In the labor market,GAI induces profound changes through three interlocking mechanisms:substitution,comple⁃mentarity,and creation.The substitution effect is no longer limited to low-skilled roles;it is increasingly affecting highskilled,white-collar occupations,thus threatening broad swaths of the workforce.The complementarity effect offers op⁃portunities to augment human productivity by offloading sub-tasks to intelligent systems,thereby raising the skill ceiling for workers.The creation effect facilitates the emergence of entirely new tasks and occupations,offering the potential for reemployment and sectoral transitions.Nonetheless,these gains are counterbalanced by the displacement of specific roles,such as those in writing,programming,and customer support,raising risks of structural unemployment and in⁃come polarization.The net outcome depends heavily on the pace of technological diffusion,the responsiveness of educa⁃tional and training systems,and the design and implementation of forward-looking labor policies.Beyond the economic domain,GAI generates far-reaching societal risks,including threats to data privacy,increased market concentration,proliferation of disinformation,and the erosion of ethical boundaries.Collectively,these chal⁃lenges have been termed the“Turing Trap”,which describes the risk of AI development deviating from its intended pur⁃pose of enhancing human welfare.To realize a“Turing Transformation”,a multidimensional governance framework is needed-one that centers on equitable technology access,human-centered design principles,and institutional safeguards to ensure that GAI evolves along a pro-human trajectory.As GAI’s influence deepens,several critical research challenges remain unresolved.These include explaining the macro-micro productivity gap,understanding long-term structural shifts in labor markets,and developing robust,multidi⁃mensional evaluation tools to assess GAI’s impacts on welfare and inequality.Future scholarship should aim to bridge micro-level behavior with macroeconomic outcomes,refine accounting systems,enhance empirical tracking methods,and promote interdisciplinary collaboration.These efforts are essential to unravel the complex socioeconomic conse⁃quences of GAI and inform the institutional design necessary for managing its long-term effects.
作者
汪红驹
丁少斌
WANG Hongju;Ding Shaobin(University of Chinese Academy of Social Sciences,Beijing,China)
出处
《经济学动态》
北大核心
2025年第8期191-208,共18页
Economic Perspectives
基金
国家社会科学基金“一带一路”建设研究专项课题(19VDL015)
中国社会科学院智库基础课题(ZKJC240903)。
关键词
生成式人工智能
图灵陷阱
图灵变换
技术革命
Generative AI
Turing Trap
Turing Transformation
Technological Revolution
