A new research paper presents a framework for measuring how artificial intelligence is affecting employment, introducing a metric called 'observed exposure.' This measure combines theoretical large language model (LLM) capabilities with real-world usage data, emphasizing automated and work-related applications. The study finds that AI’s actual use remains far below its theoretical potential, with only a fraction of feasible tasks currently automated. Occupations with higher observed exposure are projected by the U.S. Bureau of Labor Statistics to grow more slowly through 2034.
The analysis shows that workers in highly exposed professions tend to be older, female, more educated, and higher-paid. Despite these exposure levels, researchers find no systematic increase in unemployment among these workers since late 2022. However, there is some evidence that hiring of younger workers has slowed in occupations with higher AI exposure.
The authors emphasize that while AI’s labor market effects remain limited so far, their framework provides a foundation for tracking economic changes as AI adoption expands and capabilities advance.
Key Findings We introduce a new measure of AI displacement risk, observed exposure, that combines theoretical LLM capability and real-world usage data, weighting automated (rather than augmentative) and work-related uses more heavily AI is far from
The ‘1 Nojor’ media platform is now live in beta, inviting users to explore and provide feedback as we continue to refine the experience.
