人工智能如何重塑劳动力市场：创造还是替代？

王文涛，修博文

（重庆理工大学 经济金融学院，重庆 400054）

摘要：随着科技的迅猛发展，人工智能对劳动力市场的影响日益受到关注。基于2011—2023年沪深A股上市公司数据，通过文本挖掘与机器学习的方法构建企业人工智能指标，从微观企业层面研究人工智能如何影响劳动力就业。研究结果显示，人工智能在总体上促进了劳动力就业，主要表现为对高技能劳动力需求的增加和对低技能劳动力需求的减少。在影响机制方面，人工智能通过提高企业平均薪酬、优化企业学历结构和推动产品市场增长来影响劳动力就业结构。异质性分析结果表明，在低研发背景的董事会的企业、低技术行业和低要素市场发展水平的地区，人工智能对劳动力需求的影响更加明显。上述发现拓展了人工智能就业效应的研究范畴，为优化人工智能背景下的高质量充分就业政策提供决策依据。

关键词：人工智能；劳动力就业；高质量充分就业；创造效应；替代效应

作者简介：王文涛（1988—），男，重庆理工大学经济金融学院助理研究员；修博文（2000—），男，重庆理工大学经济金融学院硕士研究生。

引用格式：王文涛,修博文.人工智能如何重塑劳动力市场:创造还是替代?[J].首都经济贸易大学学报,2025,27(6):67-81.

How Artificial Intelligence is Reshaping the Labor Market: Creation or Substitution?

WANG Wentao, XIU Bowen

(Chongqing University of Technology, Chongqing 400054)

Abstract: With the rapid advancement of artificial intelligence (AI) technology, its impact on the labor market has drawn increasing attention. Using panel data from Chinese A-share listed companies from 2011 to 2023, this paper employs text mining and machine learning techniques to construct firm-level AI indicators, empirically investigating how AI reshapes employment patterns from the perspective of enterprises.

The results show that AI overall promotes employment, reflected in increased demand for high-skilled labor and decreased demand for low-skilled labor. Specifically, AI affects labord emand through raising average wages, optimizing the educational structure of the workforce, and expanding product market scale. These mechanisms jointly drive the creation effect of high-skilled jobs while exerting a substitution effect on low-skilled positions. Heterogeneity analyses reveal that the impact of AI on labor demand varies across firms with different governance structures, industries with varying technological intensities, and regions with differing levels of factor market development. Notably, the substitution effect on low-skilled labor is particularly pronounced in firms with low R&D-intensive boards, non-manufacturing firms, non-labor-intensive enterprises, and regions with underdeveloped market factors. The findings expand the research scope of AI employment impact, providing decision-making insights for optimizing high-quality full employment policies in the AI era.

This paper makes the following contributions. First, drawing on the theoretical framework of Acemoglu and Restrepo (2018), this paper incorporates AI as a capital-biased factor into the simplified production function and constructs a theoretical model of AI's impact on labor employment. Second, based on annual reports of listed companies, this paper constructs firm-level AI indicators by refining and processing text data, employing a foundational AI keyword list and utilizing the Word2Vec algorithm to dynamically generate an AI keyword set, thereby improving the scientific rigor and applicability of micro-level AI measurement. Third, from the perspective of enterprises, this paper examines the impact of AI on labor employment and investigates the key mechanisms. By analyzing the heterogeneity across firm, industry, and regional levels, this paper enriches the theoretical and empirical understanding of AI's employment impact, providing policy guidance for promoting high-quality, sufficient employment in the context of AI adoption.

This paper proposes the following policy implications. First, the government should increase investment in vocational and digital skills training programs for low-skilled workers to mitigate the risk of displacement. Second, enterprises should be encouraged to enhance recruitment and training of high-skilled talent, supported by tax incentives and talent subsidies. Third, regional policies should be tailored to local economic and industrial characteristics to promote coordinated regional development.

Keywords: AI; labor employment; high-quality full employment; creation effect; replacement effect

下载全文