AI-related skills are clustering in Chinese cities while platformization widens precarity: algorithm- and data-intensive vacancies concentrate in urban hubs, routine jobs face higher exposure to substitution, and workers often misperceive or misalign their skills with employer demand, producing regional and demographic divides.

Main Finding

AI adoption in China is reshaping labor demand and worker experience in divergent ways: platform-based work expands opportunity access but reduces stability and increases fear of technological substitution, while employers concentrate premium demand for algorithm-intensive skills in urban hubs. These dynamics produce occupational polarization (routine jobs more exposed; high-skill jobs adapt better), regional and demographic divides, and persistent perceptual and skills mismatches that require policy intervention and extensions to task-based labor models that incorporate bidirectional feedbacks.

Key Points

• Dual-perspective approach: the paper analyzes both employee-side responses (adaptation, skill portfolios, perceived risk) and employer-side demand (hiring signals, skill premium).
• Platform employment: sustains income and work opportunities but erodes job stability and raises substitution anxiety among workers; platformization contributes to precarious forms of employment.
• Occupational heterogeneity and polarization: routine and repetitive occupations show higher exposure to displacement risk; high-skill roles display greater capacity to adapt or complement AI.
• Employer stratification: firms disproportionately seek algorithm- and data-intensive skills in urban clusters, creating geographic concentration of high-value AI jobs.
• Perceptual and skill mismatches: workers’ risk perceptions, skill acquisition, and employer hiring needs are misaligned, producing frictions in reallocation and upskilling.
• Conceptual contribution: advances task-based models by incorporating bidirectional feedbacks between worker beliefs/behavior and employer demand, rather than one-way technology → tasks mapping.
• Ethical and distributional concerns: youth unemployment, an ageing population, and structural fragility mean vulnerable groups face amplified risks; unequal access to training and regional job clusters can worsen inequality.
• Policy direction emphasized: transparent regulation of AI, targeted upskilling, and equity-focused governance to mitigate displacement and distributional harm.

Data & Methods

• Data sources:
  • 2023 China Social Survey (nationally representative) — used to measure worker adaptation, skills, employment type (platform vs. traditional), and perceived AI-related risk.
  • 2023 AI recruitment big data — scraped/collected employer job postings to characterize demand for algorithm- and data-related skills and geographic clustering of vacancies.
• Analytical strategy (dual-perspective/task-based):
  • Employee-side analysis: descriptive and multivariate analyses linking employment type, occupational routines, and skills to perceived substitution risk and adaptation behaviors.
  • Employer-side analysis: text/skill extraction from recruitment data to identify premium skills, sectoral and spatial concentration of demand, and stratification across firm types.
  • Integration: combine survey and recruitment evidence to document supply–demand mismatches and feed insights back into an extended task-based framework that models feedback loops (worker expectations → upskilling behavior → employer demand).
• Methods likely include skill extraction/natural language processing on job ads, occupational routine indices, and regression controls for demographic and regional covariates (paper emphasizes nationally representative and large-scale recruitment data to support inference).

Implications for AI Economics

• Modeling: AI labor-economics models should incorporate bidirectional dynamics — worker perceptions and upskilling behavior can shift task supply and thus employer demand, producing endogenous feedbacks and path dependence in labor-market outcomes.
• Distributional impact: AI is likely to increase wage and employment polarization across occupations, regions, and demographic groups; urban clusters benefit disproportionately due to concentrated demand for algorithmic skills.
• Policy priorities:
  • Active upskilling and reskilling programs tailored to routine-task workers and regions outside major urban clusters.
  • Strengthen social protections and reduce platform precarity (e.g., portable benefits, unemployment insurance adapted to gig work).
  • Promote equitable access to AI-related education and digital infrastructure to prevent widening regional inequalities.
  • Regulatory measures for algorithmic transparency and accountability in hiring and platform management to protect vulnerable workers and reduce discriminatory outcomes.
• Research agenda:
  • Causal identification of AI adoption effects on employment trajectories (firm-level and longitudinal studies).
  • Evaluation of targeted training interventions and geographic job-creation policies.
  • Further integration of perceptual and behavioral responses into task-based and matching models to predict dynamic labor-market outcomes under AI diffusion.