Six measurements, one interpretation · 2026 Edition
Key findings
Every statement below is computed from the published dataset; the same figures appear in the county table and the downloads. Measurement first, then the author's reading of it — labeled as interpretation.
-
Higher-overlap work accounts for a larger share of pay
High- and substantial-exposure occupations account for 27.4% of Colorado's covered jobs but 34.8% of its modeled payroll (using national average wages), and the payroll share exceeds the job share in all 64 counties — exposure skews toward higher-wage work. The gap is widest in Broomfield County (35.3% of jobs; 44.0% of modeled payroll).
-
The highest shares are in Front Range metro counties
The five counties with the largest share of jobs in high- or substantial-exposure occupations — Broomfield (35.3%), Denver (33.1%), Boulder (32.6%), Arapahoe (31.0%), and Douglas (30.8%) — are all Front Range metro counties, and together they hold 44.9% of the state's UI-covered employment.
Broomfield County · Denver County · Boulder County · Arapahoe County · Douglas County
-
County shares vary, but the estimates vary in strength too
The share of jobs in high- or substantial-exposure occupations runs from 15.9% in Gilpin County to 35.3% in Broomfield County — more than double, though Gilpin's occupation mix is the most heavily estimated in the atlas (86.4% of its employment in industry cells suppressed in the public data). The 0–100 mean exposure score, kept as a secondary measure, spans a proportionally narrower range: 24.7 (Morgan) to 35.2 (Broomfield).
-
In observed Claude use, assistance leads automation
In a one-week sample of consumer Claude.ai conversations (August 4–11, 2025; Anthropic Economic Index), model-classified Colorado usage leans toward augmentation over automation — 53.9% to 46.1%, a stronger augmentation lean than the 50.9%-to-49.1% U.S. baseline.
-
Some counties have much less local data
In 17 of 64 counties, more than half of employment sits in industry cells suppressed in the public data, so their occupation mixes — and scores — are the most heavily estimated in the atlas. Together these counties hold 1.0% of state employment; the imputed share peaks at 86.4% in Gilpin County.
Interpretation
What the numbers suggest
The author's reading of the figures — labeled as interpretation, distinct from the measured data above.
What the numbers suggest
This section is the author’s interpretation of the atlas’s estimates. Occupations in the top two exposure tiers — scores of 45 or above, based on human task ratings made in 2023 — hold an estimated 27.4 percent of Colorado’s UI-covered jobs and 34.8 percent of its modeled payroll. Payroll here uses national average wages, not observed Colorado pay; see methodology limitation 10. On that basis, jobs in the top two tiers carry above-average modeled wages. The contrast describes the employment mix, not wages at risk of disappearing.
The estimates are geographically concentrated. Broomfield’s 35.3 percent is the highest county share, and eight of the thirteen counties in the top band are Front Range metros. At the other end, Gilpin’s 15.9 percent is based on 86.4 percent imputed employment — the atlas’s highest imputed share. Pueblo sits at 20.6 percent, below the typical county. These comparisons describe rated task overlap; they do not establish that a low-share county faces less economic hardship or that change will happen there later.
Job counts add another perspective. Denver, Arapahoe, El Paso, Jefferson, and Boulder together hold 64.0 percent of the estimated 784,367 jobs in the top two tiers. Including Broomfield, Douglas, and Larimer raises that share to 77.1 percent. Five small counties also appear in the top band: Bent, Crowley, Dolores, Mineral, and Park. Together they hold 1,638 jobs in the top two tiers, about 0.2 percent of the statewide total; four of the five have more than half their underlying employment imputed. A high county share and a large number of jobs are different facts.
My reading is that Front Range employers and workers deserve substantial attention in any investigation of how AI is changing work. The concentration makes them useful places to study adoption, changing duties, and hiring. That is a proposal about where to investigate, not a finding about where displacement begins. Rural access to training, local alternatives, and observed worker needs should also shape a response.
The atlas does not establish whether these occupations will shrink, expand, or reorganize. Its narrower contribution is a public baseline showing where rated task overlap is concentrated in Colorado’s estimated employment mix. Understanding what happens next requires evidence about actual workplaces.
Interpretation by the author; figures from this atlas’s dataset (2026 edition, v1.2).