There is still much ambiguity about AI’s evolution and its future impact on workers and labor markets. In Bharat Chandar’s (Stanford Digital Economy Lab) post, “AI and Labor Markets: What We Know About Don’t Know About AI and Labor Markets,” he notes that the “mismatch between supply and demand for research in this topic is unlike anything I have seen.” Here are three new measures moving us forward in understanding AI’s effect on labor markets.
Anthropic’s Observed Exposure
In March 2026, Anthropic introduced a new measure called observed exposure, which quantifies: “of those tasks that LLMs could theoretically speed up, which are actually seeing automated usage in professional settings?”
Anthropic uses the combined data from the O*Net occupation profiles database (U.S. Department of Labor), their own Anthropic Economic Index, and estimates from the research paper, “GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models” by Eloundou et al. (2023), which measure whether it is theoretically possible for an LLM to make a task at least twice as fast.
From the findings, the ten most exposed occupations include:
- Computer programming
- Customer service representative
- Data entry keyers
- Medical record specialists
- Market research analysts and marketing specialists
- Sales representative, wholesale and manufacturing, except technical and scientific products
- Financial and investment analysts
- Software quality assurance analysts and testers
- Information security analysts
- Computer user support specialists
Overall they found find no impact on unemployment rates for workers in the most exposed occupations, however, in these areas, early evidence suggests that hiring has slowed slightly for workers aged 22-25.
Coface and the Observatory of Threatened and Emerging Jobs (OEM) Joint Detailed Mapping Study
A new joint study from Coface and the Observatory of Threatened and Emerging Jobs (OEM) provides a detailed mapping of the domains where AI is most likely to reshape work. The granularity offered by breaking down 923 professions into tasks uncovers vulnerabilities overlooked when using broader labor statistics. Key findings include:
- More than 25% of work content could be automated in management and administration, creative professions, law and finance, as well as engineering and IT.
- Face‑to‑face services, technical roles, crafts, and industrial production remain below 10%. “The human dimension continues to provide strong protection.”
- Jobs in care, education, sales, and other people‑focused professions fall in between.
- Advanced economies with a strong focus information-intensive jobs, such as in the UK, the Netherlands, Ireland, and Luxembourg have higher exposure levels.
Stanford Digital Economy Lab’s Research Findings
The Stanford Digital Economy Lab’s “Canaries in the Coal Mine” paper (October 2025) uses “high-frequency” administrative data from U.S. payroll software provider, ADP, to examine changes in the labor market for occupations exposed to generative AI. Six key facts are presented:
- In AI-exposed occupations, younger workers (ages 22-25) experienced 16% relative employment declines (this mirrors Anthropic’s similar finding)
- Employment for experienced workers remained stable
- Adjustments occur primarily through employment rather than compensation
- Employment declines are concentrated in occupations where AI is more likely to automate, rather than augment, human labor
- Results exclude technology firms and remote occupations, but findings are still robust
Image by Gerd Altmann from Pixabay
May 8, 2026
