Substitution and Enhancement: Transforming the Labor Market

The potential impact of AI on the labor market has sparked intense debate. Some researchers predict that AI will lead to mass unemployment, whereas others argue that the economic prosperity brought about by AI will create new employment opportunities. While discussions of AI’s ultimate impact tend to drift into science fiction, focusing our inquiry on AI’s impact on employment for the foreseeable future—say, the next 10 years—allows for economic analysis that can provide valuable insights. Although AI has demonstrated extraordinary capabilities in some domains, current AI technology still falls short of surpassing the upper limit of human intelligence, and its development roadmap does not diminish the value and responsibility of human intelligence. At present, before further disruptive technological breakthroughs arise, the differences and complementary aspects between AI and human intelligence suggest potential for collaboration. Drawing on large-scale job posting data from China, we estimate the extent to which various occupations have been impacted by AI. AI has both substitution and enhancement effects on labor. Occupations in areas such as office administration, transportation and logistics, and data processing are at higher risk of being replaced by AI, while those in the areas of sales, legal, and management are enhanced by the technology. Based on our calculations, AI may lead to a slowdown in overall employment growth over the next decade, but mass unemployment is unlikely. We also analyze the impact of AI on wage inequality using recruitment big data from China. The results show that occupations with higher AI substitution effects have seen slower wage growth over the past few years, thereby widening the wage gap relative to occupations subject to lower substitution effects. Over the next five years, AI could lead to a modest decline in China’s labor income share. We further analyze the impact of AI on human capital, and find that the market value of education and work experience may change in response to AI. Notably, the massive amount of popular content generated by AI may reduce the market value of human work and return on human capital, potentially constraining the development of human intelligence. In the face of labor market transformation brought about by AI, policies need to address and handle both the primary distribution of income and its redistribution. The primary distribution stage should focus on vocational training and labor protection to improve workers’ ability to collaborate with AI, promoting employment and raising labor income with minimal market distortion. Redistribution that favors workers can ensure that AI development remains Pareto-improving, fostering public support for technological progress. Redistribution should be funded through highly progressive income tax with minimal distortions. In contrast, a “robot tax” may discourage investment in technology and should not be the first option. Universal basic income (UBI), which has been widely discussed in developed countries, could be a redistributive mechanism in the AI era, but it entails high fiscal costs. A more pragmatic approach for China is to improve the existing social security system, especially by leveraging AI-driven growth to enhance the support and protection of vulnerable groups.