The heterogeneous effects of artificial intelligence on labor markets: A calibrated simulation of skills, tasks, and wages

Artificial Intelligence (AI) is reconfiguring labor markets with profound, heterogeneous effects on skills and wages. While existing models offer qualitative insights, they often lack the quantitative rigor for testable predictions. This paper develops a novel simulation model extending the standard task-based framework from a one-dimensional continuum to a two-dimensional space (cognitive complexity and codifiability) with a dynamic AI capability frontier. Calibrated to match empirical stylized facts, such as the high-skill wage premium, our model generates three key findings. First, AI's advance consistently widens the wage gap between high- and low-skill workers. Second, the rate of inequality growth slows when AI becomes a stronger substitute for human labor, as broad displacement compresses wage growth across the skill spectrum. Third, counterfactual policy simulations reveal a strict efficiency-equity trade-off: interventions managing AI's adoption pace (e.g., an "AI tax") effectively mitigate inequality but dampen aggregate output, while training subsidies and labor supply expansions prove less effective against structural automation. Our work provides a methodologically transparent tool for analyzing the future of work. • Calibrated model reveals rising wage inequality between high/low skills. • Inequality growth slows when AI acts as a strong labor substitute. • Low-skill wages stagnate while high-skill wages rise with AI expansion. • AI tax reduces inequality 4× more effectively than training subsidies. • Provides open-source Python code for replicable AI-labor market analysis.