Artificial Intelligence and Occupational Risk: Evidence from a Longitudinal Panel of Six Economies

College of Computer Science and Technology, China University of Petroleum (East China), QingDao, 266500, China

^* This email address is being protected from spambots. You need JavaScript enabled to view it.

Abstract

What determines how exposed a worker is to AI substitution, and does that exposure translate directly into lower pay. To address these questions, the paper draws on the Future of Jobs AI Dataset—a panel of 12,343 worker-year observations across 10 occupations in six countries (2015–2035)—and apply a combination of OLS regression with fixed effects, K-means cluster analysis, and two machine learning models. Over the 21-year window, the mean ai_risk_score climbed from 0.316 to 0.467, a 47.9% rise that points toward slow-moving structural change rather than a short-run shock. Regression estimates indicate that the primary technical skill a worker holds matters far more than years of experience or educational attainment: relative to Python specialists, Excel users face an ai_risk_score premium of 0.447 points, and each additional unit of risk is associated with roughly $61,419 less in annual salary. The Random Forest classifier recovers AI risk categories with 83.4% accuracy on held-out data, while a Gradient Boosting model explains 95.5% of salary variance—both results reinforcing the OLS pattern. Cluster analysis groups the ten occupations into three bands whose risk and wage profiles diverge sharply, suggesting that targeted reskilling toward Python, cloud, and deep learning competencies could yield measurable wage gains.