Adaptation · Labour & consumers
Employment in AI-exposed occupations, year on year
Year-on-year employment change in the most AI-exposed occupation quintile versus the least exposed, from the Stanford DEL and ADP Research Canaries Dashboard (payroll data). The band applies to the most-exposed series.
Reason through thisThe reading
Exposed occupations are flat while unexposed grow slowly; the signal is in the young, not the aggregate.
50 readings drawn from canaries.us_exposure_q5.employment_yoy.m, one of 5 series listed under How it is measured; 103 observations in the evidence.
Source: Canaries Dashboard data (by exposure). Stanford Digital Economy Lab and ADP Research, AI Economic Indicators: Canaries Dashboard.
The numbers
| as of | reading | value | record |
|---|---|---|---|
| 2026-09-30 | us exposure q5 | -0.5% | reported |
| 2026-08-31 | us exposure q5 | -0.4% | reported |
| 2026-07-31 | us exposure q5 | 0.4% | reported |
| 2026-06-30 | us exposure q5 | 0.2% | reported |
| 2026-05-31 | us exposure q5 | -0.1% | reported |
| 2026-04-30 | us exposure q5 | 0.4% | reported |
| 2026-03-31 | us exposure q5 | 0.3% | reported |
| 2026-02-28 | us exposure q5 | 0.4% | reported |
| 2026-01-31 | us exposure q5 | 0.4% | reported |
| 2025-12-31 | us exposure q5 | 0.3% | reported |
| 2025-11-30 | us exposure q5 | 0.3% | reported |
| 2025-10-31 | us exposure q5 | 0.7% | reported |
| 2025-09-30 | us exposure q5 | 0.5% | reported |
| 2025-08-31 | us exposure q5 | 0.2% | reported |
| 2025-07-31 | us exposure q5 | 0.5% | reported |
| 2025-06-30 | us exposure q5 | 0.3% | reported |
| 2025-05-31 | us exposure q5 | 0.7% | reported |
| 2025-04-30 | us exposure q5 | 0.7% | reported |
| 2025-03-31 | us exposure q5 | 0.8% | reported |
| 2025-02-28 | us exposure q5 | 1.1% | reported |
| 2025-01-31 | us exposure q5 | 1.3% | reported |
| 2024-12-31 | us exposure q5 | 1.9% | reported |
| 2024-11-30 | us exposure q5 | 1.6% | reported |
| 2024-10-31 | us exposure q5 | 1.6% | reported |
| 2024-09-30 | us exposure q5 | 1.5% | reported |
| 2024-08-31 | us exposure q5 | 2.3% | reported |
| 2024-07-31 | us exposure q5 | 1.7% | reported |
| 2024-06-30 | us exposure q5 | 1.8% | reported |
| 2024-05-31 | us exposure q5 | 2.3% | reported |
| 2024-04-30 | us exposure q5 | 2.1% | reported |
| 2024-03-31 | us exposure q5 | 2.5% | reported |
| 2024-02-29 | us exposure q5 | 2.7% | reported |
| 2024-01-31 | us exposure q5 | 2.7% | reported |
| 2023-12-31 | us exposure q5 | 2.5% | reported |
| 2023-11-30 | us exposure q5 | 3.2% | reported |
| 2023-10-31 | us exposure q5 | 3.2% | reported |
| 2023-09-30 | us exposure q5 | 3.7% | reported |
| 2023-08-31 | us exposure q5 | 4.0% | reported |
| 2023-07-31 | us exposure q5 | 4.1% | reported |
| 2023-06-30 | us exposure q5 | 5.4% | reported |
| 2023-05-31 | us exposure q5 | 5.7% | reported |
| 2023-04-30 | us exposure q5 | 5.6% | reported |
| 2023-03-31 | us exposure q5 | 6.1% | reported |
| 2023-02-28 | us exposure q5 | 6.2% | reported |
| 2023-01-31 | us exposure q5 | 6.6% | reported |
| 2022-12-31 | us exposure q5 | 7.3% | reported |
| 2022-11-30 | us exposure q5 | 7.8% | reported |
| 2022-10-31 | us exposure q5 | 8.2% | reported |
| 2022-09-30 | us exposure q5 | 8.8% | reported |
| 2022-08-31 | us exposure q5 | 8.5% | reported |
Why it matters
The paper predicts no broad AI-attributable employment decline for years; this is the highest-frequency payroll test of that claim.
What cuts against it
Exposure is a model-derived measure; ADP payroll coverage skews to larger firms; the divergence began before ChatGPT on some cuts (NY Fed; Iscenko and Millet).
How it is measured
- series
canaries.us_exposure_q5.employment_yoy.m - series
canaries.us_exposure_q1.employment_yoy.m - series
adp_research.us_high_exposure.employment_yoy.m - series
adp_research.us_low_exposure.employment_yoy.m - series
fortune.us_high_exposure.employment_yoy.m - source ADP Research, Canaries Dashboard releases · default tier 6 · ADP Research; short quotation
- source Canaries Dashboard data (by exposure) · default tier 6 · Stanford Digital Economy Lab; free public data, cited as the Lab asks
- source Fortune · default tier 5 · Fortune Media; short quotation
- proxy deployment, welfare · cadence monthly · valve adoption to adaptation
Normal = within one percent of flat, ordinary occupational churn; fast = a 3% or worse decline in exposed occupations, which is what a broad AI-attributable displacement looks like in payroll data. Between is `emerging`.
Applied to canaries.us_exposure_q5.employment_yoy.m.
Latest readings
Why this status, and its history
Evaluator: -0.002 (ADP Research, Canaries Dashboard releases, 2026-06-30) is inside the consistent band (lo=-0.01 hi=0.01). Auto-reason; band rationale: Normal = within one percent of flat, ordinary occupational churn; fast = a 3% or worse decline in exposed occupations, which is what a broad AI-attributable displacement looks like in payroll data. Between is `emerging`.
- 2026-09-10unmeasured to consistent with normalconf — → 80 · the evaluator (automatic)
Evaluator: -0.002 (ADP Research, Canaries Dashboard releases, 2026-06-30) is inside the consistent band (lo=-0.01 hi=0.01). Auto-reason; band rationale: Normal = within one percent of flat, ordinary occupational churn; fast = a 3% or worse decline in exposed occupations, which is what a broad AI-attributable displacement looks like in payroll data. Between is `emerging`.
source1
Why this confidence
Rests on 103 observations from ADP Research, Canaries Dashboard releases, Canaries Dashboard data (by exposure), Fortune; best evidence tier 5, grade B; current.
Confidence is independent of status: 90–95 multiple strong independent sources; 70–89 good evidence, some ambiguity; 50–69 mixed or hard to operationalise; 0–49 limited or vague evidence.
Predictions that lean on this
- Too early to tellAI does not wipe out software engineering jobs, and demand for engineers may rise as software gets cheaper to make. · Arvind Narayanan and Sayash Kapoor
- Too early to tell