人工智能、任务重构与技术红利的分配

Artificial intelligence (AI) not only improves production efficiency but also reshapes the structure of job tasks and the distribution of income between labor and capital. AI can both substitute for existing tasks and create new ones, so different forms of task restructuring may have divergent effects on productivity and income distribution. Identifying the channels through which AI generates productivity gains and affects their distribution is therefore essential for reconciling technological progress with employment stability and income equality.

Drawing on the annual reports of China’s A-share listed companies and data from online recruitment platforms, this paper uses text analysis and task matching to measure firm-level AI adoption and job-level task substitution, task expansion, and posted wages. Using an instrumental-variable approach, it examines AI’s productivity effects, task-restructuring mechanisms, and distributional consequences. The results show that AI increases firm TFP through both task substitution and task expansion, although the two channels have opposing distributional effects. Task expansion increases posted wages and the labor income share of non-executive employees, whereas task substitution significantly reduces both. Because the negative effect of task substitution on labor income outweighs the positive effect of task expansion, AI-generated productivity gains disproportionately accrue to capital. These effects are more pronounced in service-sector firms, which experience both larger productivity gains and a greater decline in the labor income share of non-executive employees.

This paper develops measures that link firm-level AI adoption to job-level task restructuring. It explains both the generation and distribution of AI-generated productivity gains by distinguishing between the channels of task substitution and task expansion, thereby providing new micro-level empirical evidence on the economic effects of AI. The findings suggest that policymakers should improve education and vocational training systems, encourage firms to adopt human–AI collaboration and task-expanding applications, and strengthen labor protection and benefit-sharing mechanisms to enable workers to share more fully in the productivity gains generated by AI.