Does artificial intelligence affect the pattern of skill demand? Evidence from Chinese manufacturing firms

In view of the recent penetration of artificial intelligence (AI) into production activities, we undertake a quasi-natural experiment to identify its impact on employment at different skill levels using micro-enterprise data from Chinese manufacturing during 2011-2017. Employing a robust difference-in-differences method with propensity score matching, we investigate the heterogeneous impact of AI adoption upon different skills across three dimensions - geographical regions, enterprise types, and the length of time since the adoption of AI. We find that AI reduces the relative demand for low-skilled labor across all regions in China, while increasing the relative demand for high-skilled labor only in the eastern region. These differential impacts of AI upon relative demand for different skills reflect firm-level technological intensity. Results also show that the longer the duration of AI use, the greater is the impact upon the relative demand for high-skilled labor.