Commentary: How AI Is Quietly Fixing a Key Problem in China’s Labor Market

A recent study by Peking University’s National School of Development AI and Economics Lab and the recruitment platform Zhaopin examines how artificial intelligence—particularly large language models and generative AI—are transforming China’s labor market, based on extensive data from job postings and applications[para. 1]. The researchers found persistent and even deepening structural mismatches between worker education/skills and job requirements—situations known as vertical and horizontal mismatch[para. 2]. Vertical mismatch refers to overqualified individuals taking lower-skilled jobs, while horizontal mismatch means a person’s field of study does not match their role[para. 2]. The study reveals that about 40% to 45% of job applications showed either type of mismatch, and the rate of mismatch among "near-hires" (where employer and candidate interactions are especially strong) has sharply increased in recent years, suggesting that even more considered candidate-job pairings are increasingly mismatched[para. 3].

The researchers attribute these persistent mismatches to the expansion of higher education outpacing the transformation of economic structures, resulting in a surplus of highly educated job seekers forced to "downgrade" their job searches to less suitable positions[para. 4]. However, a turning point appears around late 2022, with the launch of ChatGPT acting as a landmark for an intensified AI "shock." In the wake of this event, high-AI-exposure occupations—such as technology R&D, data analysis, and content creation—began to experience a decrease in both overqualified applicants and mismatched near-hires. That is, matching in these high-tech fields has started to improve[para. 5].

Despite a surge in applications for AI-related jobs after ChatGPT’s release, companies became more selective—employer and positive response rates fell as organizations faced more candidates but higher volumes of unqualified applicants. Thus, the post-ChatGPT labor market requires more precise screening and matching[para. 6]. Several mechanisms help explain how AI brings these improvements[para. 7]. First, enhanced job postings (with more specific, detailed, and explicit requirements) create clearer signals for job seekers, which reduces irrelevant applications and increases matching efficiency[para. 8]. Second, as AI transforms job content, the entry barrier is systematically raised—roles demand more complex human-computer collaboration and advanced skills, prompting self-selection by candidates who better understand the requirements[para. 9]. Third, AI drives greater specialization in roles and task descriptions, leading to a more targeted labor-market fit[para. 10].

The study concludes that AI’s disruptive force may actually improve labor market efficiency and partially resolve structural mismatches—at least in sectors most affected by AI—by pushing companies toward more precise talent screening and job design[para. 11]. However, this positive effect is largely confined to high-AI-exposure jobs for now; across the broader labor market, educational mismatch is worsening, with continued risks of excessive competition among educated candidates and an overall decline in job quality[para. 12].

Accordingly, the researchers recommend forward-looking policies: enterprises should be supported in leveraging technology to improve job matching, while education and vocational training must evolve more rapidly to align human capital with new labor-market demands[para. 13]. This study is based on Zhaopin’s job data from 2018-2025, covering about 100,000 jobs annually, and their corresponding applicant records[para. 14].

[para. 1]