Glossary
Key terms and acronyms used throughout AI Exposure.
Augmentation
When AI handles specific tasks within a job while the human worker focuses on higher-value activities. Contrasts with displacement, where the entire role is eliminated.
Automation Probability
The estimated likelihood (0–100%) that a given occupation's tasks can be performed by machines or software. Originally computed by Frey & Osborne (2017) for 702 occupations using machine learning classifiers.
BLS
Bureau of Labor Statistics — the principal federal agency responsible for measuring labor market activity, working conditions, and price changes in the US economy.
Composite Risk Score
AI Exposure's proprietary 0–100 score combining five weighted factors: Frey/Osborne probability (30%), OECD task analysis (20%), BLS projections (20%), layoff signals (15%), and GenAI exposure (15%).
Displacement
When automation eliminates an entire job or role, requiring the worker to find employment in a different occupation. Contrasts with augmentation.
FRED
Federal Reserve Economic Data — a database maintained by the Federal Reserve Bank of St. Louis containing 800,000+ US and international economic time series.
GenAI (Generative AI)
AI systems that create new content — text, images, code, audio, video — based on patterns learned from training data. Includes large language models (ChatGPT), image generators (DALL-E, Midjourney), and code assistants (GitHub Copilot).
GenAI Exposure Index
A measure of how much an occupation's tasks overlap with capabilities of large language models and generative AI tools. Based on the Eloundou et al. (2023) methodology.
Gig Economy
A labor market characterized by short-term, freelance, or contract work rather than permanent employment. Includes platforms like Uber, DoorDash, Fiverr, and Upwork.
ISCO
International Standard Classification of Occupations — the ILO's international equivalent of SOC codes, used for cross-country comparisons of occupational data.
LLM (Large Language Model)
A type of AI model trained on massive text datasets to understand and generate human language. Examples: GPT-4, Claude, Gemini, LLaMA. The core technology behind most generative AI text applications.
NAICS
North American Industry Classification System — the standard used by federal agencies to classify business establishments by industry. Uses 2–6 digit codes (e.g., 52 = Finance and Insurance).
Non-Routine Task
A job task requiring flexibility, creativity, problem-solving, or interpersonal skills that are difficult to codify in algorithms. Examples: negotiation, creative writing, emergency response.
O*NET
Occupational Information Network — a free database from the US Department of Labor with detailed descriptions of tasks, skills, abilities, and work contexts for ~1,000 occupations.
OEWS
Occupational Employment and Wage Statistics — a BLS survey producing annual employment and wage estimates for 800+ occupations across all industries and geographic areas.
Routine Task
A job task that follows explicit rules and can be accomplished by machines following well-defined procedures. Can be manual (assembly line) or cognitive (data entry, bookkeeping).
RPA (Robotic Process Automation)
Software tools that automate repetitive, rule-based digital tasks like data entry, form filling, and report generation. Popular tools include UiPath, Automation Anywhere, and Blue Prism.
Skills Overlap
A percentage (0–100%) measuring how much of the skills required for one occupation transfer to another. Used to identify viable career transition paths.
SOC Code
Standard Occupational Classification code — a 6-digit identifier (e.g., 15-1252 for Software Developers) used by federal agencies to classify workers into occupational categories for consistent data collection.
Task-Based Analysis
An approach to studying automation that evaluates individual tasks within a job rather than the job as a whole. Recognizes that most occupations contain a mix of automatable and non-automatable tasks.
Transition Path
A related occupation with lower automation risk and high skills overlap, representing a viable career pivot. AI Exposure calculates transition paths based on O*NET skill and task similarity scores.
WARN Act
Worker Adjustment and Retraining Notification Act — federal law requiring employers with 100+ employees to give 60-day advance notice of plant closings or mass layoffs affecting 50+ workers.