Career Training Path › Research › AI exposure

AI exposure by trade

On the AI Occupational Exposure index, construction laborers rank in the bottom 4% of all occupations for exposure to AI capabilities, massage therapists the bottom 7%, and welders, cutters, solderers, and brazers the bottom 12%. For comparison, accountants and auditors sit at the 99th percentile and financial managers at the 98th.

Exposure measures how much of an occupation's required abilities overlap with what current AI systems can do. It is not a prediction of job loss: some exposed work gets augmented, and hands-on work that AI cannot physically perform scores low even where paperwork around it gets automated. Read it alongside pay and openings, and as one input among several.

All program groups

Office jobs, for comparison

Method

AIOE links 10 AI application capabilities (language modeling, image recognition, translation, and so on) to the 52 O*NET abilities each occupation requires, and scores every occupation on how much of its ability profile those capabilities touch. Scores are standardized: 0 is the average occupation, positive is more exposed, negative less. Exposure measures overlap with what AI can do, not job loss — some exposed work is augmented rather than replaced. Percentile is the share of all scored occupations with lower exposure. Source: AI Occupational Exposure index (AIOE), Felten, Raj and Seamans; BLS Occupational Employment and Wage Statistics May 2025. Occupation codes are matched from the 2018 Standard Occupational Classification to the 2010 codes the index uses where they differ.