AI exposure

basicwidespread

AI exposure shows what proportion of a job's tasks artificial intelligence can perform or significantly accelerate.

In plain terms

AI exposure is a metric that estimates how much of your work artificial intelligence can touch. It does not look at the job title, but at daily tasks: you write emails, enter data, negotiate, fit pipes, lift patients.

It is like a flood map. It shows which streets the water might reach, but not whether your house will get wet: that depends on the levee, the house foundation, and what you do in time. High exposure similarly only indicates that your work will be reached by AI. It does not mean it will take it away.

Researchers therefore distinguish two cases. In one, AI augments: it takes the routine off your shoulders, and you do more and better. In the other, it replaces: it solves the task without you. The same high exposure can thus bring wage increases in one profession and layoffs in another.

Unlike previous waves of mechanization, it is not physical but mainly office and intellectual work that is most exposed: writing, analysis, programming, and administration.

How it works

Researchers break down occupations into tasks, mostly based on large occupational databases: for example, an accountant's job consists of dozens of separate activities.
For every task, it is estimated whether current AI can perform it or at least significantly accelerate it. Experts or the AI models themselves score this, and increasingly, it is measured from real usage data.
Task scores are aggregated, so every occupation receives an exposure value, for instance, what percentage of its tasks are affected.
Some analyses examine whether AI in a given job complements or replaces humans, as this determines whether exposure represents an opportunity or a threat.
The values are finally compared with employment data, making it possible to calculate for countries, sectors, genders, and education groups who is most affected.

An everyday example

You read an article stating that 'a significant portion of Hungarian jobs are exposed to AI,' and you get scared because you work in customer service. The figure actually means that AI can speed up most of your tasks (replying to letters, summaries, data searches). Whether this results in an easier workday or staff reductions depends on your employer's decisions and your own adaptability.

Who is affected

employeeFrom exposure data, you can see which of your tasks AI can take over or speed up, and you can adjust what is worth learning accordingly. → Work & HR

jobseeker, new graduateIn high-exposure professions, AI often first reaches entry-level, routine tasks, so it is worth considering this when choosing a career. → Work & HR

entrepreneur, executiveMapping exposure shows where AI can bring efficiency to the company and which employees need retraining.

parent, student, teacherIt helps determine which skills remain valuable even when machines perform routine intellectual tasks. → Education & Family

A common misconception

MythIf a profession is heavily exposed to AI, it will cease to exist.

In realityExposure only means that AI can touch a large share of the tasks. Many exposed jobs are transformed or become more productive, and only some of them see a reduction in headcount.

Limits and risks

Exposure measures technical possibility, not a forecast: it does not account for costs, regulations, liability issues, or how slowly companies adopt new technologies.
Methods differ, so different studies on the same profession can yield very different figures. Always check who measured and how.
Measurement becomes outdated quickly: a task that was not exposed to AI last year may be performed this year by a new model or an autonomous AI agent.
The average hides much: in the same profession, an experienced professional may gain from AI, while the beginner's routine tasks disappear.

Similar, but not the same

Automation risk – Automation risk estimates the chance that a machine will replace a human, while exposure only indicates that AI affects the work, even as a helper.
AI adoption rate – Adoption measures how many people actually use AI, while exposure measures what AI could be used for.
Technological unemployment – This refers to jobs actually disappearing due to the technology, while exposure is only a preliminary estimate that does not necessarily lead to unemployment.

Related concepts

Broader

The impact of artificial intelligence on the labor market, Automation, The future of work

Narrower

Task-based exposure, Complementary AI, Replacement AI

On the edge

Automation risk, Reskilling, Productivity growth, Generative AI, AI-agent

A brief history

The idea's precursor is Carl Benedikt Frey and Michael Osborne's 2013 Oxford study, which estimated the automatability of occupations. AI-specific exposure metrics emerged around 2018, such as the AIOE index by Edward Felten and co-authors. The term gained widespread use after the launch of ChatGPT in 2023, when OpenAI researchers and university co-authors assessed the impact of large language models by occupation. In January 2024, an International Monetary Fund analysis made it globally known: its estimate suggests nearly 40 percent of jobs worldwide are exposed to AI, and roughly 60 percent in advanced economies.

Where things stand

By the end of 2026, exposure has become a standard metric in labor market analyses, used regularly by international organizations, central banks, and governments. The focus is shifting from theoretical estimates to real usage data and measuring actual employment impacts.

Status: 2026 November

In the news

Appears in 5 Agent Age stories; the curve shows weekly occurrence in the news magazine.

Where in the US the most jobs are exposed to AIOct 20, 2026 · Work & HR
Studies: AI exposure alone does not show where employment grows or shrinksOct 5, 2026 · Work & HR
Forecasts: McKinsey says automation will affect 36 million jobs by 2035Oct 5, 2026 · Work & HR
Cognizant to hire 1,500 graduates while putting AI exposure at $4.5tnSep 7, 2026 · Work & HR
Employment rising in Denmark's jobs most exposed to AI, says Arbejderbevægelsens ErhvervsrådAug 10, 2026 · Work & HR

Also known as

AI exposure · AI exposure · AI exposure · AI exposure · AI occupational exposure · occupational exposure to AI · AIOE · exposure to generative AI

Sources

Eloundou, Manning, Mishkin, Rock: GPTs are GPTs – An Early Look at the Labor Market Impact Potential of Large Language Models (2023)

This entry was machine-edited and has undergone machine expert review; the “In the news” section updates automatically from Agent Age stories.