Apollo Research Links AI Exposure to Slower Wages
Apollo Global Management published research on July 30 finding that US occupations with the highest AI exposure experienced an average 6.7% decline in real wage growth after 2023, while employment showed no detectable change. The analysis matched 321 occupations and used Anthropic's Claude usage-based Economic Index as its AI-exposure measure, according to Apollo and Business Insider.
Apollo Global Management published a July 30 analysis finding that US occupations with the highest exposure to AI experienced an average 6.7% decline in real wage growth after 2023, while employment levels in those occupations showed no detectable change.
Apollo Chief Economist Torsten Slok and analyst Sania Edlich wrote that the research used a difference-in-differences design with occupation and year fixed effects across 321 matched occupations from 2015 onward. Apollo's Daily Spark described the exposure measure as based on actual Claude usage data and characterized the result as slower wage growth in AI-exposed occupations rather than lower employment.
Business Insider reported that the study combined occupational and wage data from the US Bureau of Labor Statistics with Anthropic's Economic Index. That index estimates occupational AI exposure from the share of related tasks observed being performed with Anthropic's AI tools.
Wage effects varied across workers
According to Business Insider's account of the paper, service workers recorded a 24.3% decline in earnings growth since 2023, while workers in the bottom quartile of earners saw wages decline by 10.7% over the period. The publication reported that the paper found no statistically significant effect among the highest-paid workers.
Business Insider also listed examples from Apollo's comparison of 2022 and 2024 BLS data, including a 6.1% real-wage decline for computer programmers and a 5.4% decline for statistical assistants. The available article excerpt did not provide the complete occupation-level results.
Interpreting an exposure-based result
The paper acknowledges data challenges, including changes to occupational and industry classifications and data collection over time, Business Insider reported. Those limitations matter because an AI-exposure index captures observed task use, not a direct measure of whether a particular employer deployed AI across an entire role.
For ML and data practitioners, the study adds labor-market evidence to a debate often framed solely around job displacement. Research designs that pair tool-usage data with occupational outcomes can surface important distributional patterns, but exposure measures and observational wage data do not by themselves establish that AI caused every measured wage change. Replication with additional model-use datasets, employer-level adoption data, and longer post-adoption periods would help test the reported relationship.
Key Points
- 1Apollo's 321-occupation analysis links high AI exposure with 6.7% lower real wage growth after 2023, without detectable employment effects.
- 2Business Insider reports larger wage-growth declines among service and lower-paid workers, while the paper found no significant effect for top earners.
- 3Observed AI tool usage offers a new labor-market measurement input, but broader datasets are needed to distinguish exposure correlations from causal effects.
Scoring Rationale
The research presents a notable empirical claim about AI's labor-market effects using occupational wage data and a Claude usage-based exposure index. It is relevant to AI practitioners and policymakers, although the reported findings are observational and come from a single analysis rather than a new technical system or broad regulatory action.
Sources
Primary source and supporting public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
