Artificial Intelligence and Heterogeneous Unemployment Risk Across Regions: Scenario-Based Projections of Alternative Policy Responses in Taiwan.

OBJECTIVE: To evaluate the impact of artificial intelligence (AI) on employment in Taiwan by quantifying exposure and projecting unemployment risks across industries, occupations, and regions. METHODS: National workplace survey data (N = 4,009) were analyzed using AI Industry Exposure (AIIE) and Occupation Exposure (AIOE) indices to construct a composite AIECI. Six scenarios (α = 0.075 or 0.15; RA = 0, 0.5, 1) modeled unemployment projections for 2025-2035. RESULTS: Taipei, Hsinchu, and Taichung showed the highest exposure. Under high-impact scenarios, urban unemployment may rose sharply, while retraining interventions reduced projected risks. Rural regions remained less affected. CONCLUSIONS: AI exposure is unevenly distributed, concentrating risk in technology-intensive regions and occupations. Targeted workforce adaptation policies are needed to mitigate unemployment and regional disparities.