High vs Low AI-Exposure Worker Characteristics
Data
Per massenkoff-mccrory-labor-market-impacts-2026 Figure 5. US Current Population Survey, Aug–Oct 2022 (immediately before ChatGPT release).
Comparison groups: Top quartile of observed-exposure workers (N=32,301) vs zero-exposure workers (N=42,546).
| Dimension | No exposure | Top quartile | Difference |
|---|---|---|---|
| AI coverage % | 0.0% | 38.8% | +38.8 pp |
| Demographics | |||
| Age (years) | 41.0 | 42.9 | +1.9 |
| Female | 38.8% | 54.4% | +15.5 pp |
| Hispanic | 24.8% | 13.8% | −11.0 pp |
| White, non-Hispanic | 54.5% | 65.1% | +10.6 pp |
| Black, non-Hispanic | 13.2% | 9.7% | −3.5 pp |
| Asian, non-Hispanic | 4.7% | 9.1% | +4.4 pp |
| Married | 44.6% | 54.9% | +10.4 pp |
| Education | |||
| Less than HS | 13.2% | 2.3% | −10.9 pp |
| HS diploma | 38.9% | 17.7% | −21.2 pp |
| Some college / assoc. | 30.0% | 25.5% | −4.6 pp |
| Bachelor’s | 13.3% | 37.1% | +23.8 pp |
| Graduate degree | 4.5% | 17.4% | +12.8 pp |
| Labor market | |||
| Hours / week | 37.5 | 38.7 | +1.2 |
| Hourly wage ($) | $22.23 | $32.69 | +$10.45 (+47%) |
| Union member | 11.7% | 5.3% | −6.4 pp |
| Observations | 42,546 | 32,301 | — |
Key asymmetries
- Graduate degree share is ~4× larger in exposed cohort (17.4% vs 4.5%).
- Bachelor’s share nearly triples (37.1% vs 13.3%).
- 47% wage premium for exposed work.
- Halving of union membership (5.3% vs 11.7%) — exposed work is overwhelmingly non-unionised white-collar work.
- 15.5 pp more female — but conditional on the education profile, not the prior-bear-case “low-education clerical women” framing.
- 11 pp less Hispanic — exposure is concentrated in non-Hispanic-white and Asian cohorts.
See
- exposed-worker-demographics — interpretation and theoretical implications.
- most-exposed-occupations — the occupational composition that drives these demographics.
Citation
Figure 5, Massenkoff & McCrory (2026), p. 9. CPS Aug–Oct 2022. Top-quartile defined by employment-weighted occupation-level exposure score.
Related
- massenkoff-mccrory-labor-market-impacts-2026 — source.
- exposed-worker-demographics — concept page that interprets this table.
- observed-exposure-measure — the measure used to define quartiles.
- capital-labor-divergence — surplus-distribution lens.
- acemoglu-simple-macroeconomics — prior incidence prediction partially contradicted by this table.
- H2_u-curve-of-value — labor-side U-shape (different axis from the value-stack U).