Live· BLS never·0 forecasts
Studio

AI Exposure Scoring · Now

Compose rubrics, batch-score occupations, and route low-confidence outputs to human review.

Mean exposure (today)
51/100

Rubric editor · v1.2

Versioned · auditable · diff vs v1.1

Score distribution · Now

050100

Review queue · 15 occupations

Approve, override or leave unreviewed
OccupationSOCScoreΔ vs todayConf.RationaleStatus
Paralegals & Legal Assistants23-2011800.92Document review, summarization and citation are largely automatable.
Registered Nurses29-1141320.73Documentation, triage support and patient education are LLM-augmentable, but bedside care requires physical presence and clinical judgment.
Construction Managers11-9021410.76Scheduling, risk and bid review are model-augmentable; site coordination is not.
Waiters & Waitresses35-3031260.70Ordering kiosks expand; hospitality remains human-centered.
Executive Secretaries & Admin Assistants43-6011880.95Scheduling, drafting and inbox triage are now AI-native.
Cashiers41-2011820.93Self-checkout and contactless payments displace headcount.
Home Health & Personal Care Aides31-1120180.67Predominantly physical caregiving; AI assists scheduling and care notes only.
Financial Managers11-3031660.86Modeling, variance analysis and memo drafting are highly augmentable.
Plumbers, Pipefitters & Steamfitters47-2152120.65Physical labor in unpredictable environments.
Customer Service Representatives43-4051840.94Conversational AI handles tier-1 contact at scale.
General & Operations Managers11-1021550.82Reporting, forecasting and comms are augmentable; people decisions are not.
Retail Salespersons41-2031500.80Self-checkout and digital displacement; in-store advisory remains human.
Physicians & Surgeons29-1210480.79Diagnostic imaging, summarization, differential generation are model-augmentable; final accountability remains human.
Elementary School Teachers25-2021360.74Lesson planning and grading are augmentable; classroom management is not.
Human Resources Managers11-3121520.81Screening, policy drafting, comms are augmentable; sensitive cases need humans.
Human–AI collaboration · Now
≥50% crossover · May 2026
Tool-assisted20%
Co-pilot80%
Agent team0%
Supervised swarm0%
Agent-executed financial transactions
Now
40%
of finance $-volume agents can run unattended
Trust ceiling now: Read-only insights · cap $0
AI literacy enforcement
Now
42
/100 enforcement maturity
Active stage: DOL AI Literacy Framework
39.1M workers in scope · DOL ETA framework Feb 2026