人工智能对就业规模及劳动收入的影响——来自Meta分析的证据

The Effects of Artificial Intelligence on Employment Scale and Labor Income--Evidence from a Meta-Analysis

采用Meta分析方法,基于近年来28篇相关文献的40个独立样本,分析人工智能对就业规模与劳动收入的总体影响,并进一步将人工智能分为表征内生的自身投入和外生的外部引入两个维度进行探讨,运用Meta亚组分析探究潜在调节变量对上述关系的调节作用。研究表明:总体上看,人工智能小幅扩大了就业规模,效应值为0.032;降低了劳动收入,效应值为-0.011。具体而言,技术的投入维度显著地促进就业规模和劳动收入,而使用维度则对二者起到抑制作用。此外,本文识别了行业、地区作为调节变量的影响,研究结果显示人工智能对制造业、发展中国家就业规模的提升作用大于非制造业和发达国家;人工智能增加了制造业和发展中国家的劳动收入,但降低了发达国家的劳动收入。

In recent years,the application space of artificial intelligence has become more and more extensive.As a representative technology of digital economy,its impact on economic society and people's livelihood has been concerned by scholars.One of the most controversial is the impact of artificial intelligence on employment.At present,the research conclusions on the relationship between artificial intelligence and employment scale and labor income are quite different.Revealing the effect of artificial intelligence on employment scale and labor income is of guiding significance for objectively assessing the impact of artificial intelligence on the labor market and actively coping with it.Therefore,this paper attempts to use meta-analysis method to explore the relationship between artificial intelligence and employment scale and labor income,and to use this method to analyze existing empirical studies in a more comprehensive and systematic way.At the same time,existing literatures are summarized and differences are sorted out.Specifically,based on 40 independent samples of 28 relevant literatures in recent years,this paper analyzes the overall impact of artificial intelligence on employment scale and labor income,and further divides artificial intelligence technology into two subdivision dimensions,namely endogenous technology input and exogenous technology use,in order to draw more accurate research conclusions.Meanwhile,by summarizing and analyzing previous studies,it is found that industry and region may be potential moderating variables among the main variables studied in this paper.Therefore,Meta subgroup analysis is used to explore the moderating effects of the two on the above relationship.The study shows that:overall,AI has slightly expanded the size of employment,with a correlation coefficient of 0.032,and reduced labor income,with a correlation coefficient of-0.011.Specifically,the input dimension of technology significantly promotes employment scale and labor income,while the use dimension inhibits both.This conclusion provides empirical evidence for the input innovation of enterprises.In addition,this paper identifies the effects of industry and region as moderating variables,and finds that AI has a greater effect on employment in manufacturing and developing countries than in non-manufacturing and developed countries.Artificial intelligence has increased labor incomes in manufacturing and developing countries,but reduced labor incomes in developed countries.In general,from the perspective of AI subdivision dimension and the identification of moderating variables,this study found the reasons for the general and inconsistent conclusions in previous studies,and provided empirical basis and practical enlightenment for the impact of different AI development levels on the labor market,including more scientific research conclusions and policy recommendations for reference for the study of the relationship between the two.At the same time,there are three research limitations in this paper.Therefore,in the future research,it is hoped that the majority of scholars can conduct more in-depth exploration of the impact of artificial intelligence and employment,so that people can have a more comprehensive understanding of artificial intelligence and it simpact.Thus,artificial intelligence,the most representative general technology in the fourth industrial revolution,can be integrated into social life more smoothly and efficiently,and enable human development.