Every number is cited and reproducible.
The Government console runs the same scoring engine as JobRoute Individuals and Companies, with no per-persona modifications. This page documents the four pillars, the 11-archetype classifier, the pipeline-gap proxy, and all six data sources. Same input always produces the same score.
How to read these scores.
The AI Ready Score is a measurement, not a prediction. A low score signals that workers in an occupation face elevated AI exposure today, based on publicly available task and skill data. Policy responses (retraining programs, apprenticeships, pipeline investments) can change outcomes. The score is a starting point for evidence-based planning, not a verdict on workers or occupations.
Where a jurisdiction's employment distribution is synthetic (derived from national BLS data scaled to state share), the public transparency view says so explicitly. The data provenance field in every briefing records whether figures are real or modeled.
Four scoring pillars.
Overall score = (Task Exposure inverted × 0.35) + (Skill Durability × 0.25) + (Role Trajectory × 0.20) + (Adjacency Breadth × 0.20). Range: 0 to 100. Higher = more AI-ready.
Task Exposure
Weighted average of per-task AI-usage exposure (from the Anthropic Economic Index, or a documented keyword and occupation-category heuristic where AEI has no direct observation) times O*NET task importance, inverted so that lower exposure yields a higher score.
How much do current AI systems get used on the tasks in this occupation? Workers in high-exposure roles face the most immediate displacement pressure.
Skill Durability
A category-based durability rating for the occupation skill profile, adjusted by WEF Future of Jobs trajectory signals. Skill-posting-decline data is a planned enhancement.
Do the skills in this occupation hold their value as AI spreads? Caregiving, critical thinking, and creativity score high.
Role Trajectory
BLS 10-year projected employment change, mapped to a 0-100 score around a neutral baseline, with a modest stability adjustment for larger occupations.
Is overall demand for this role growing or contracting?
Adjacency Breadth
Count of reachable adjacent occupations within a 12-month skill gap, normalized on a log scale.
How many adjacent paths exist for workers in this role? High breadth means more retraining options and lower dislocation risk.
11-archetype classifier.
Each occupation is assigned one of 11 archetypes by half-open interval rules over Task Exposure, Skill Durability, and Adjacency Breadth bands, with Euclidean nearest-prototype tie-breaking. Fully deterministic: same input always produces the same archetype. Urgency levels (Stable, Watch, Act Now) drive the emerald, amber, and vermillion color scale throughout the console.
Pipeline gap.
Proxy caveat
The pipeline-gap metric uses BLS employment counts in high, mid, and low exposure bands as its input, not age or career-stage data (which state labor market data does not consistently provide). The Stanford/ADP study coefficients are applied as per-band employment delta percentages. This is an honest approximation: the output is stated as a modeled estimate, not a direct measurement.
Formula: weighted sum of (employment in band × Stanford/ADP coefficient for that band) / total employment. Coefficients from the "Canaries in the Coal Mine" study, Stanford Digital Economy Lab and ADP Research Institute, November 2025.
Six data sources.
Task definitions and importance ratings for 1,016 occupations. Primary source for task-level AI exposure assessment.
Per-task observed AI-usage exposure (a measure of how much AI is used on a task, not an automation probability; augmentative use is weighted below full automation). Where AEI has no direct observation for a task, exposure is estimated by a documented keyword and occupation-category heuristic. Combined with O*NET task importance to compute the Task Exposure pillar.
Employment totals by occupation. BLS publishes employment at the 6-digit SOC level; where a SOC maps to several O*NET occupations, the real SOC total is distributed evenly across them, so occupations sharing a SOC show an equal allocated share rather than a separately measured count. Jurisdiction totals still sum to the real state figure. Jurisdictions without ingested state OEWS data use national figures scaled to state share and are labeled modeled in every briefing and public view.
Skill durability ratings. Skills rated as human-resilient are weighted as durable for the Skill Durability pillar.
"Canaries in the Coal Mine" study: employment delta percentages by AI-exposure band and early-career cohort. Coefficients power the Pipeline Gap metric.
130,402 occupation-pair edges derived from O*NET skill overlap. Powers the Adjacency Breadth pillar and adjacent-role recommendations.