More info
Glossary
A–Z of dashboard terms, from AIOE to WARN.
Every acronym and metric ID used across the dashboard is defined here — no need to go hunting through the methodology pages to figure out what CM2 or L4 means.
A
- Adoption
- The share of businesses that say they use AI, from the Census Bureau's Business Trends and Outlook Survey. Published in two eras with different question wordings, which are never merged into one line.
- AI exposure score
- A measure of how much of a job's task time AI is actually used for, based on published AI-usage data. Shown relative to the national average (1.00 = average, 1.20 = 20% above).
- AI occupation exposure
- How much AI-assisted work happens in a specific type of job, measured from real AI usage rather than predicted. A higher score means more of the job's task time involves AI.
- Augmentation
- When AI helps a worker do their job better or faster rather than replacing them. The worker remains employed but uses AI as a tool.
B
- BLS projection
- The Bureau of Labor Statistics' 10-year forecast of how many jobs an occupation will gain or lose. A positive projection means the field is expected to grow.
C
- Canaries
- The Stanford Digital Economy Lab / ADP "Canaries" work, which tracks employment at young firms in the occupations most exposed to AI — an early-warning signal for displacement.
- CBSA
- Core-Based Statistical Area — the government's term for a metro area. It groups together a city and its surrounding communities that are economically connected.
- Composite index
- A single score built by combining several different measurements. The local stress score is the plain average of unemployment, the job-postings index, and layoff notices (at least two must be current, or it is withheld).
D
- Displacement
- When a worker loses their job because their tasks can be done by AI or automation, and there are not enough new jobs in their field to absorb them.
H
- Heuristic
- An estimate based on rules of thumb rather than direct measurement. When we do not have a published study for a job, we use heuristics to approximate its AI exposure.
J
- Job postings index
- A measure of how many new job openings employers are advertising compared to a baseline. A declining index means employers are hiring less.
- JOLTS
- Job Openings and Labor Turnover Survey — a government report that tracks how many jobs are open, hired, and quit each month. A decline in openings can signal a cooling labor market.
M
- Metro stress
- A measure of labor-market pressure in a metro area, combining unemployment, declining job postings, and layoff notices. Higher stress means the local job market is struggling.
N
- National AI-Impact Index
- The AI-weighted composite: mean over ranked metros of (stress_score × clamp(0.83425 + 1.8 × (exposure − 0.6685), 0.5, 1.0)), where exposure is the metro AI-exposure weight (0-1) and the anchors are fixed constants per DEC-04 (superseding the earlier raw 0.5 + 0.5 × exposure form). Below 30 is low, 30-60 is moderate, 60+ is high. It is served at the API layer and powers the National Pulse impact layer — no longer headlined on the home page (per DEC-32).
- National Risk Index
- The unweighted mean of stress_score across all tracked US metro areas (0-100 scale). Below 30 is low, 30-60 is moderate, 60+ is high. This is the number shown in the Quick Report National Context block.
O
- Occupational exposure
- How much a specific type of job is affected by AI. Jobs with high occupational exposure have many tasks that AI can perform.
P
- Pivot path
- An alternative career that uses similar skills to your current job but has lower AI risk. Pivot paths show you safer directions to transition into.
- Planning region
- Connecticut's county-equivalent geography — the state now reports its statistics for nine planning regions rather than counties, and dashboard data for Connecticut uses them.
- Provenance
- The record of where a number came from and what was done to it — its source class (official, private, research, or computed) and its transform (raw, derived, interpolated, imputed, or modeled).
- PUMA
- Public Use Microdata Area — a geographic zone of about 100,000 people used by the Census Bureau for detailed survey data. Smaller than a metro area but larger than a neighborhood.
Q
- QCEW
- Quarterly Census of Employment and Wages — a government dataset that counts jobs and wages by county and industry every three months. It is the most accurate local employment data available.
S
- Sahm Rule
- A recession indicator that triggers when the unemployment rate rises by half a percentage point above its recent low. It is one of the earliest reliable signs of a downturn.
- Skill overlap
- How much two jobs share similar skills and tasks. High overlap means switching between them is easier because you already have most of the needed abilities.
- Suppressed value
- A value the source agency withholds, usually to protect confidentiality. Suppressed values are shown as missing, never as zero.
V
- Vintage
- The publication date attached to a data value. When a value is re-published on a new day, the new value is stored as a new vintage alongside the old one — history is never overwritten.
W
- WARN notice
- A legally required notice that an employer must give 60 days before a mass layoff. It is an early warning sign of job losses in your area.
- WARN notices
- Legally required notices that employers must give 60 days before mass layoffs. They are early warning signs of job losses in your area.