Core Insight: Artificial intelligence serves as the core engine of the new technological revolution and industrial transformation, and is a decisive factor for China to seize the global industrial high ground. However, given that the productivity surge brought by AI can easily become exclusive gains for capital and data factors, it is recommended that relevant ministries actively monitor the impact of AI applications on employment and income distribution. Establishing supporting mechanisms such as data dividends, skill enhancement programs, and flexible regulatory frameworks can gradually and appropriately transform AI's "efficiency dividends" into "income dividends" for workers, achieving a dual victory in technological leadership and common prosperity.
China's recently released 14th Five-Year Plan proposes the comprehensive implementation of the "AI Plus" initiative, strengthening the integration of AI with technological innovation, industrial development, cultural development, social welfare, and governance. The plan aims to capture the high ground in AI industrial applications and empower all sectors of the economy. Since 2018, breakthroughs in pre-trained large models like ChatGPT, Sora, and DeepSeek have driven widespread penetration of AI technology across economic and social domains. While significantly enhancing corporate competitiveness, AI is also reshaping factor distribution patterns, presenting new challenges for protecting workers' rights.
I. AI Drives Deep Restructuring of Employment Landscape, Increasing Structural Imbalance Risks From a macro perspective, structural employment contradictions are prominent. High-skill positions, complementary to AI, continue to increase in number. Although low-skill roles relying on human interaction are difficult to fully replace, the transfer of workers from medium-skill positions to low-skill fields has caused a surge in job seekers and intensified competition in these areas. Monitoring data from 300 Chinese cities by the Chinese Academy of Labor and Social Security shows that from January to August 2025, demand for AI-related positions such as algorithm engineers, AI trainers, and data analysts grew by over 100% year-on-year, while demand for sales, administrative, financial, and legal positions declined by 10% to 30%.
Labor's share of income is declining. AI improves production efficiency, but new corporate revenues are primarily used for purchasing intelligent equipment, software systems, accumulating data assets, and expanding investments or dividends, rather than significantly raising employee wages. For example, Tencent's 2024 revenue was 665.8 billion yuan, with employee compensation at 59.2 billion yuan, accounting for 8.9% of revenue—significantly lower than the 11.2% in 2019. Compared to hundred-billion-level capital investments in AI computing power, smart production lines, and data systems, labor's share in new value distribution is clearly insufficient.
Regional and industrial development imbalances are worsening. AI applications heavily depend on data infrastructure, high-skilled talent, and capital investment, causing technological dividends to concentrate more in digitally developed regions and leading industries. Research indicates that the proportions of provinces in eastern, central, western, and northeastern China with AI industrial competitiveness above the national average are 4/5, 2/3, 1/4, and 1/3, respectively, showing a declining gradient from east to west. The "technology geography agglomeration" effect leaves regions with dense traditional manufacturing or lower service digitization relatively lagging in job absorption capacity and industrial upgrading speed.
From a micro perspective, job competency requirements are rising. While AI replaces basic tasks like data organization and report writing, companies simultaneously raise performance standards, increasing demands for quick response, comprehensive judgment, and cross-domain integration capabilities. Higher complexity work significantly raises actual job thresholds. Research from Management World shows that occupations with higher AI exposure experience more significant demand contraction, intensified internal salary disparities, and potentially higher educational and experience requirements for hiring.
Actual working hours are increasing. Algorithmic management in the AI era decomposes labor into a state of "continuous responsibility, discrete time." Under fragmented on-call mechanisms, workers' 24-hour responsibility severely diverges from recorded working hours, resulting in statistically lower but actually higher working hours. Stanford University research indicates that while global AI adoption exceeded 87% in 2025, average working hours per employee increased by 1.5 hours, with overtime rates in AI-assisted positions 34% higher than in traditional roles.
Career development paths are narrowing. As basic positions are simplified or replaced by AI, workers lose traditional promotion paths through experience accumulation, falling into a structural dilemma of "having positions but no progression." Tesla's Shanghai factory, through "lights-out" automated production, achieved a 75% robot replacement rate, increasing weekly capacity to 5,000 vehicles per plant with 60% fewer workers than traditional factories. Numerous assembly line operator positions have disappeared, replaced by a few technicians monitoring and maintaining equipment, eliminating the former path of promotion through years of frontline experience.
II. AI Increases Efficiency but Not Relief: Three Root Causes of Labor Time Dilemmas From an individual perspective, AI causes "task fragmentation," weakening workers' perception of "real marginal output" and creating greater uncertainty in labor value recognition and compensation returns. Neoclassical labor supply theory suggests that wage rate increases produce income and substitution effects. When AI raises marginal hourly wages, the income effect leads workers to desire more leisure and fewer hours, while the substitution effect prompts them to substitute work for leisure, extending working hours. The net result depends on the perception of "marginal benefit." If marginal benefits are unpredictable, the income effect may be delayed or disappear, leaving the substitution effect dominant. Unable to confirm if "more work" yields "more gain," workers hesitate to reduce hours, while the substitution effect, reinforced by algorithmic performance systems, remains dominant. Systems continuously assign new tasks and intensify quantitative KPIs, forcing workers to passively extend working hours to meet baseline requirements.
From an enterprise perspective, the failure of AI-driven productivity gains to translate into social leisure stems from labor market power asymmetry and weakened collective bargaining power. AI's impact on working hours essentially reproduces production organization logic and power relationships. Companies prioritize using marginal output growth from AI for business expansion, market share increases, and shareholder returns. Under the dominant paradigm of "maximizing shareholder value," management lacks incentive to convert efficiency gains into social leisure. Simultaneously, the rise of the gig economy exacerbates worker atomization, severely weakening collective bargaining power against massive platform algorithms. Even with increased individual productivity from AI, workers lack sufficient negotiation leverage to demand "reduced hours and increased pay."
From an institutional perspective, labor time regulations lag behind technological change, lacking effective constraints on AI algorithms. Traditional labor law systems are based on industrial-era logic of "fixed hours + clear employer responsibility." However, new AI-driven work models (e.g., platform gig work, remote smart collaboration, human-machine hybrid task flows) blur boundaries between "work" and "on-call," and "production time" and "preparation time." Though not physically "on duty," workers remain in psychological readiness due to real-time task dispatch, immediate response requirements, and invisible online assessments, resulting in actual working hours far exceeding legal limits.
Therefore, AI's impact on working hours depends not only on technological attributes but also on power relationships within economic institutions. Efficiency gains from the substitution effect could provide a material basis for shortening working hours, but under current production organization logic, these benefits are primarily captured by enterprises and transmitted as higher labor intensity through task restructuring, performance pressures, and market competition. While freed from repetitive tasks, workers face more complex "cognitive involution," potentially reducing their actual welfare. Thus, market mechanisms alone cannot ensure a positive correlation between technological progress and human well-being; institutional optimization is necessary to reshape benefit distribution patterns.
III. Policy Recommendations As the core engine of the new technological revolution and industrial transformation, and a decisive factor for China's global industrial leadership, AI requires careful management. Given that its productivity surge easily becomes exclusive capital-data gains, relevant ministries should actively monitor AI's impact on employment and income distribution. Establishing tools like data dividends, skill enhancement, and flexible regulation can gradually and appropriately convert AI's "efficiency dividends" into "income dividends" for workers, achieving technological advancement and common prosperity.
First, guided by "shared development," explore reasonable distribution mechanisms for AI efficiency dividends. Upholding "development for the people," encourage qualified regions and industries to pilot AI efficiency-sharing initiatives while protecting corporate innovation incentives. Support enterprises in using portions of cost savings from AI applications to establish "digital skills enhancement special funds," determining fund usage through collective consultation, focusing on skills training, career transition, and development for affected employees. Additionally, promote listed companies' disclosure of "AI's impact on labor input and countermeasures" in ESG reports, enhancing social supervision and transparency to ensure technological benefits are fairly shared with workers.
Second, focus on "skill leapfrogging" to build an integrated "prediction-training-employment" support system. Leveraging the Ministry of Human Resources and Social Security's "Skills China Action," accelerate the establishment of dynamic monitoring and response mechanisms for occupational changes under AI influence. Using recruitment platform, social security, and tax big data, quarterly release "Early Warning Lists for Job Substitution and Emerging Demand in Key Industries" to precisely identify "skill gap red lists." Based on this, promote vocational schools and technical institutes partnering with leading enterprises to create "AI collaboration talent order classes," offering "tuition-free, subsidized, guaranteed placement" full-cycle support for transitioning workers. Simultaneously, unemployment insurance funds could establish AI transition subsidies, providing living allowances for workers obtaining certification in high-demand skills, ensuring "continuous income during transition and skill advancement."
Third, premised on "safeguarding baseline welfare," promote flexible governance and humanistic care in algorithmic employment. Adhering to inclusive and prudent principles, focus on high-risk scenarios without interfering normal business operations. Human resources departments, jointly with trade unions and industry associations, should formulate "Healthy Working Hour Guidelines for AI Collaboration Positions," encouraging platforms to develop humanized features like "continuous operation reminders" and "fatigue self-assessments." For high-frequency algorithm-managed sectors like delivery, warehousing, and customer service, conduct voluntary working hour load assessment pilots, forming "key monitoring lists." For scheduling systems with systemic overtime or insufficient rest, promote optimization through union negotiations, industry self-regulation, or public explanations, ensuring workers in the AI era possess both efficiency and dignity.
Comments