Opinion: Beware AI’s ‘Dumbbell Effect’ on Jobs

The impact of artificial intelligence (AI) on employment is attracting widespread attention, driven by significant changes in the job market. According to a Zhaopin report, job postings in the AI sector in China increased by 11% year-on-year in the third quarter of 2025, while the number of job seekers in this field rose by 23%. Over the first three quarters of 2025, the AI sector experienced a 3% increase in job postings and a much larger 39% increase in the number of job seekers, indicating heightened interest and competition in AI-related employment opportunities [para. 1].

The societal response to AI’s impact on labor is split between optimism and anxiety. Some people welcome the shift, labeling critics as Luddites, while others worry about AI’s potential to cause widespread job loss and societal disruption. Though historical technological revolutions have both created and destroyed jobs, experts note that AI’s shock to the labor market is more profound than previous revolutions, increasing uncertainty and anxiety about the future. The crucial challenge now is to analyze AI’s structural impact on jobs rather than relying on speculation or fear. Understanding patterns such as which types of jobs are at risk or likely to grow is essential for adapting to these changes [para. 2].

Recent studies shed light on AI’s complex impact on labor markets. Research from Peking University highlights that generative AI, like large language models, exacerbates structural imbalances in China’s labor market. This results in a polarization where both high-skill and low-skill job opportunities increase, but mid-skill jobs decline. Similarly, Harvard University research shows that in the United States, generative AI displays a “seniority bias”: junior positions are more affected, primarily through reduced new hiring rather than layoffs, while promotion rates for current junior employees increase. The Harvard study also found that AI’s impact on job seekers is greatest for those with mid-tier educational backgrounds—graduates from second- and third-tier institutions—while those from top- or lower-tier institutions are less affected [para. 3].

The mechanisms by which AI affects employment are shaped by factors such as economic development, industrial structure, and demographics, leading to uneven impacts across sectors and regions [para. 4]. Researchers describe the simultaneous growth of high- and low-skill jobs with the decline of mid-skill roles as the “dumbbell effect.” High demand exists for interdisciplinary professionals such as algorithm engineers and AI product managers, while positions involving repetitive, standardized tasks, like administrative clerks or paralegals, are shrinking. On the other hand, low-skill roles such as data annotation provide a transition path for workers leaving traditional manufacturing sectors [para. 5].

Governments, businesses, and educational institutions must adapt to these new realities [para. 6]. Chinese higher education is reforming rapidly, with universities introducing AI-related majors and interdisciplinary programs. Workers and students alike are upskilling to boost employability in an AI-driven economy [para. 7]. The Chinese government’s 15th Five-Year Plan (2026-2030) emphasizes coordinated talent cultivation, restructuring higher education, and boosting AI literacy by mentioning “artificial intelligence” eight times in its guidelines for reform [para. 7].

To address the risks of a dumbbell-shaped job market, expanding the social safety net is vital. Some countries have considered policies such as a “robot tax” or “technology replacement tax.” China must remain alert to the dangers of a widening digital divide, turning challenges into opportunities through proactive adaptation and innovation. Ultimately, employment security in the AI era will rely on the creation of new work opportunities and a resilient, adaptable workforce [para. 8].