关键词: 人工智能, 就业质量, 就业公平

Abstract: As a core technology driving technological revolution and industrial transformation, artificial intelligence (AI) is profoundly reshaping the labor market. While existing research has extensively examined its impact on employment quantity, studies focusing on employment quality remain relatively limited. Given the breakthrough advances in AI technologies and their deepening penetration across industries, a systematic investigation into how AI development affects workers' employment quality carries significant theoretical and practical relevance,particularly for advancing the policy goal of “promoting high-quality and full employment.”

In this context, this study aims to clarify the impact of AI development on workers' overall employment quality and its sub-dimensions, explore the underlying mechanisms, and examine heterogeneity across different groups. To address these issues, this study combines city-level AI patent data with individual-level data from the China Family Panel Studies (CFPS), constructing a comprehensive employment quality index at the worker level and an AI patent density indicator at the city level. A fixed-effects model is employed for baseline regression analysis. To address potential endogeneity concerns, this study further applies an instrumental variable approach, the Heckman two-step method, and an exogenous shock identification strategy. In addition, a series of robustness checks are conducted.

The main findings are as follows. First, AI development significantly improves workers' overall employment quality： a one-standard-deviation increase in city-level AI patent density raises employment quality by 2.21%. This result holds after accounting for endogeneity and conducting multiple robustness tests. Second, analysis of the sub-dimensions of employment quality shows that AI development significantly increases labor income, improves welfare security, enhances job stability, and reduces the risk of overwork, while its effect on job satisfaction is statistically insignificant. Third, mechanism analysis indicates that AI development improves employment quality mainly by promoting occupational upgrading, strengthening human capital accumulation, and improving job-skill matching. Fourth, heterogeneity analysis indicates that the employment quality enhancement effect of AI development varies across groups. The positive effects are more prominent among workers with lower employment quality, female workers, and those with stronger non-cognitive skills.

Based on these findings, this study proposes several policy implications： actively expanding new forms of human-machine collaboration to create more high-quality jobs; building a lifelong learning system to facilitate workers' skill upgrading; and integrating non-cognitive skill development into the education system to enhance workers' comparative advantages in human-machine collaboration. Together, these measures can help achieve broad-based improvement in employment quality in the AI era. In summary, through rigorous empirical analysis, this study provides new micro-level evidence and theoretical explanations for understanding the evolution of employment quality in the AI era. It deepens theoretical insights into the role of technology in empowering workers, and offers more targeted policy implications for guiding AI development toward the promotion of high-quality employment and the creation of a more equitable labor market.

Keywords: Artificial Intelligence, Employment Quality, Employment Equity