Skill Gap Analyses

Autor and Thompson argue that standard AI exposure analyses miss a crucial dimension: whether AI automates an occupation's expert or inexpert tasks, which determines whether wages rise or fall. Using linguistic analysis of 303 occupations over four decades, they find that occupations whose expertise requirements rose by one standard deviation saw 18% wage increases but lower employment, caused by a shrinking pool of qualified workers, while those whose requirements fell saw wage declines but employment growth. They propose that skill-gap frameworks should ask not just "which jobs are exposed?" but "which tasks within each job are exposed, and are those the expert or inexpert tasks for that occupation?"

The Anthropic Economic Index provides a novel approach to measuring AI's economic impact by analyzing actual usage patterns from millions of conversations with Claude rather than relying on theoretical exposure estimates. The initial report found that roughly 36% of occupations showed AI being used for at least a quarter of their tasks. Usage remains concentrated in software development and technical writing, with a balance leaning toward augmentation (57%) over automation (43%). The January 2026 report introduced "economic primitives" to allow for more precise measurement. These standardized metrics include: task complexity, skill level (measured by education years required), purpose/use case, AI autonomy (degree of delegation), and success rate. The index also tracks geographic adoption patterns, finding that uneven worldwide adoption remains well-explained by GDP per capita, and lower-usage U.S. states are catching up relatively faster, suggesting potential equalization within two to five years.

The "Iceberg Index," is a skills-centered metric that simulates 151 million U.S. workers as autonomous agents across 32,000 skills and 3,000 counties, powered by Oak Ridge's Frontier supercomputer. The research found that current AI systems can technically perform tasks representing 11.7% of total U.S. wage value (approximately $1.2 trillion), far exceeding visible tech-sector disruption (2.2%, or $211 billion). The index challenges assumptions that AI risk is confined to coastal tech hubs; simulations show exposed occupations spread across all 50 states, including inland and rural regions. The tool “enables policymakers and business leaders to identify exposure hotspots, prioritize training and infrastructure investments, and test interventions before committing billions to implementation.”

Erik Vasilauskas and Michael Horrigan extended the Pew Research Center’s task-based framework to categorize occupations by AI exposure intensity. Using O*NET data to measure the relative importance of 16 key AI-susceptible work activities, they clustered occupations into high, middle, and low exposure tiers. Their analysis of BLS 2023–2033 projections found that high-exposure roles are concentrated in high-skill, high-education fields (e.g., analysts, professional services). Crucially, the study distinguishes outcomes within this high-exposure group: while analytical and financial roles are projected to grow via augmentation, clerical and administrative roles face decline due to direct automation.