大数据环境下人工智能的“干中学”效应研究

Research on the“Learning by Doing”Effect of Artificial Intelligence in the Digital Economy

本文对人工智能参与生产活动后其生产效率逐步提高的现象进行了学理化表达。研究表明,人工智能的机器学习能力使得其能够调用庞大数据集供自身进行训练,从而变得更加“智慧”,体现到生产活动中,人工智能具备类似于劳动力进行“干中学”的能力,本文将其称为人工智能的“干中学”效应。基于新古典增长理论的研究范式和分析框架,人工智能的“干中学”效应使得更高水平的平衡增长路径存在且稳定。同时,本文从经济奇点和要素配置等问题入手,讨论了人工智能的经济影响及其作用机制。建立在理论分析的基础上,本文选用来自中国30个省份和179个地级市的平衡面板数据进行实证检验,结果表明,人工智能的“干中学”效应对我国宏观经济总产出与全要素生产率水平产生了重要作用,但在产业、行业层面表现出一定的异质性。最后,本文从全面发挥人工智能的“干中学”效应以及推动宏观经济高质量发展的角度给出了政策建议。

This paper presents a theoretical expression of the gradual improvement in production efficiency following the involvement of artificial intelligence(AI)in production activities.The study indicates that AI’s machine learning capabilities allow it to utilize vast datasets for self-training,thereby becoming more“intelligent.”In the context of production activities,AI demonstrates an ability to learn while working,similar to human labor,which this paper refers to as AI’s“learning-by-doing”effect.Based on the research paradigm and analytical framework of neoclassical growth theory,the“learning-by-doing”effect of AI ensures the existence and stability of a higher level of equilibrium growth path.Mean while starting from issues such as economic singularity and factor allocation,this paper discusses the economic impact of AI and its mechanism of action.Founded on theoretical analysis,this paper employs balanced panel data from 30 provinces and 179 prefecture-level cities in China for empirical testing.The results show that AI’s“learning-by-doing”effect plays a significant role in China’s macroeconomic output and the level of total factor productivity,although it exhibits certain heterogeneity at the industry and sectoral levels.Finally,the paper offers policy recommendations from the perspective of fully leveraging AI’s“learning-by-doing”effect and promoting high-quality macroeconomic development.