What is AI exposure of occupations?
AI exposure is a measure of how much of an occupation's work artificial intelligence could perform or speed up. It describes the overlap between job tasks and AI capabilities, not whether those jobs will be lost.
Also known as: occupational AI exposure, AI exposure index, AI exposure score
Researched and fact-checked by AI, with no human review. 5 sources listed below. How we verify
Last updated
How it is measured
The measures described here work at the task level. Researchers break an occupation into its tasks. They judge which ones AI could perform or speed up. They combine the results into a score for the job.
A commonly used measure comes from a 2023 paper by researchers at OpenAI, OpenResearch and the University of Pennsylvania. It covers large language models, the kind of AI behind chatbots, and software built on them. It counts a task as exposed if access to either would cut the time a person needs to complete it by at least half without lowering quality. On that basis the authors estimated about 80% of U.S. workers could have at least 10% of their tasks affected. About 19% could have at least half affected, they estimated.
The International Labour Organization applies a task-based approach worldwide. Its index, updated in May 2025, sorts occupations into four rising levels of exposure to generative AI, meaning systems that produce text, images or code. It estimates one in four workers globally is in an occupation with some exposure. The estimate ranges from 11% of employment in low-income countries to 34% in high-income countries. It finds clerical occupations the most exposed.
What it does not show
Exposure is not the same as job loss. The 2023 paper states its measure does not distinguish between AI that assists workers and AI that replaces them. The International Monetary Fund estimated in January 2024 that almost 40% of global employment is exposed to AI. In advanced economies it said roughly half of exposed jobs may benefit through higher productivity, while AI could lower labor demand for the other half. The Budget Lab at Yale said in May 2026 that exposure scores are informed guesses that rest on human and AI judgment and may change or prove wrong.
Where things stand in 2026
A newer measure adds data on actual use. In March 2026 Anthropic introduced a measure it calls observed exposure. It combines theoretical capability with data on how Anthropic's Claude models are used at work. By the company's own figures, Claude covered 33% of tasks in computer and math occupations, against 94% rated theoretically feasible under the 2023 paper's measure. The same Budget Lab analysis reported no statistically significant effect so far on the employment or wages of exposed occupations. It cautioned that this could change quickly.
Sources
- GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models, arXiv (Eloundou, Manning, Mishkin and Rock)
- Generative AI and Jobs: A Refined Global Index of Occupational Exposure, International Labour Organization
- AI Will Transform the Global Economy. Let’s Make Sure It Benefits Humanity., International Monetary Fund
- What We Do and Don't Know About How AI is Affecting the Labor Market, The Budget Lab at Yale
- Labor market impacts of AI: A new measure and early evidence, Anthropic
