AI destroys low-skilled jobs while empowering high-tech talent

The study’s headline finding is clear: artificial intelligence significantly depresses the demand for low-skilled labor, while concurrently boosting employment for middle- and high-skilled workers. The impact is not evenly distributed. Low-skilled workers, often employed in repetitive, easily automatable roles, are being displaced at an accelerating pace. For every unit increase in AI technology level, demand for low-skilled labor falls by 2.047 percentage points, underscoring the potent substitution effect of AI on basic labor tasks.

China is leading the race in artificial intelligence development, but a growing body of research is uncovering deep fractures in its labor market. A recent study reveals how AI-driven innovation is fundamentally transforming employment dynamics across the country, accelerating demand for high-skilled workers, squeezing out low-skilled labor, and deepening gender and regional inequalities. With the world watching China’s AI ascent, this study offers a critical look at how rapid automation could unravel job equity if left unchecked.

The findings come from the peer-reviewed paper titled “Artificial Intelligence, Technological Innovation, and Employment Transformation for Sustainable Development: Evidence from China”, published in Sustainability (2025, 17, 3842). Authored by Hui Liang, Jingbo Fan, and Yunhan Wang, the study leverages panel data from 30 Chinese provinces between 2010 and 2022, applying rigorous fixed-effects modeling to map the evolving employment skill structure under the pressures of AI and tech-driven disruption.

How is AI changing the demand for workers?

The study’s headline finding is clear: artificial intelligence significantly depresses the demand for low-skilled labor, while concurrently boosting employment for middle- and high-skilled workers. The impact is not evenly distributed. Low-skilled workers, often employed in repetitive, easily automatable roles, are being displaced at an accelerating pace. For every unit increase in AI technology level, demand for low-skilled labor falls by 2.047 percentage points, underscoring the potent substitution effect of AI on basic labor tasks.

On the flip side, high-skilled labor sees a substantial bump, with AI developments driving a 1.174 percentage point increase in employment demand. This trend reflects the growing need for professionals who can build, manage, and collaborate with advanced AI systems. Middle-skilled workers, often assumed to be vulnerable in the automation era, also benefit in China’s context, registering a 1.192 percentage point rise in demand. This finding contradicts the job polarization theories seen in Western economies and reflects China’s unique labor composition, where middle-skilled workers are vital to maintaining operational continuity amid digital transitions.

However, the study finds a crucial nuance. While high- and middle-skilled employment grows steadily with AI advancement, low-skilled labor faces a “threshold effect.” Once AI development in a province crosses a critical point, specifically, a patent-based index value of 9.834, the decline in low-skilled job demand accelerates sharply. This non-linear collapse suggests that at a certain maturity level of AI, technologies don’t just complement or assist human labor - they fully replace it.

Who is most at risk in the AI-led labor shift?

The research reveals a worrying gender divide in AI’s employment impact. Women, particularly those in low-skilled positions, face a more severe decline in demand than their male counterparts. Low-skilled female employment is disproportionately concentrated in administrative, clerical, and assembly line work - sectors most vulnerable to automation. In contrast, low-skilled men are more likely to work in physically intensive roles like construction or logistics, which are harder to automate in the short term.

For high-skilled employment, the pattern flips. Men benefit significantly from AI-driven job creation, especially in STEM and AI development roles. Meanwhile, high-skilled female workers do not experience statistically significant gains. This imbalance reflects existing gender disparities in access to education, digital literacy, and leadership positions in the tech sector - structural issues that AI is now exacerbating rather than mitigating.

The regional picture adds another layer of inequality. AI’s impact is significantly more disruptive in labor-intensive provinces than in capital-intensive ones. In regions where industries rely heavily on low-skilled workers, such as manufacturing and light industry, AI-driven automation has decimated repetitive roles. In contrast, capital-intensive provinces, home to high-tech industries and financial hubs, experience more stable employment transitions, thanks to a larger high-skilled workforce and better institutional capacity for absorbing technological shocks.

What role does technological innovation play in labor market shifts?

A defining feature of this study is its investigation into the mediating role of technological innovation. Using measures like R&D spending and the concentration of skilled innovation talent, the authors show that innovation acts as both a trigger and a buffer for AI-induced employment shifts.

For low-skilled and middle-skilled labor, technological innovation partially mediates AI’s effects. As AI stimulates innovation ecosystems, it indirectly suppresses demand for low-skilled jobs while creating more sophisticated roles that middle-skilled workers can fill with upskilling. However, innovation plays no significant mediating role for high-skilled labor, likely because these workers are already at the forefront of AI design and deployment. Their employment gains are directly tied to AI’s core technological advancements rather than spillover effects.

This insight offers critical implications. The positive employment effects of AI are amplified where innovation capacity is high - where R&D investments, talent pipelines, and supportive ecosystems flourish. Conversely, regions or sectors lacking this innovation infrastructure see sharper job losses and fewer new opportunities. This suggests that AI doesn’t just reshape labor markets - it does so unevenly, privileging those with institutional readiness and punishing those without.

A call for inclusive and coordinated policy response

The authors argue that these disparities are not inevitable - they are a product of policy choices, or the lack thereof. To mitigate AI’s uneven impact, the study recommends a multi-pronged strategy: increased public investment in R&D, development of inclusive reskilling programs, gender-sensitive labor policies, and region-specific economic planning to cushion labor-intensive zones from disruptive shocks.

Educational reforms are also emphasized, especially the creation of interdisciplinary programs that align with AI’s emerging labor demands. From vocational retraining to lifelong learning systems, the goal is to build a resilient workforce capable of transitioning with technological change rather than being overrun by it.

The researchers stresses that failure to act could exacerbate income inequality, marginalize vulnerable groups, and destabilize the social foundations of sustainable development. Instead of a rising tide lifting all boats, unchecked AI diffusion may deepen divides - between high- and low-skilled workers, men and women, and developed and underdeveloped provinces.