How we calculate AI exposure scores
A transparent, task-based estimate of how much of a job's working time current AI can perform or speed up.
1. Break the job into tasks
For each occupation we list the five task groups that make up most of a typical working week, informed by public occupational descriptions (such as O*NET OnLine from the US Department of Labor) and practitioner input. Each task group gets a share of working time; the shares add up to 100%.
2. Rate AI capability for each task
Each task gets an AI exposure rating from 0 to 100, describing how much of that task today's widely available AI tools can perform or noticeably accelerate:
| Rating | Meaning | Typical examples |
|---|---|---|
| 0–15 | AI has little practical role | Physical work, hands-on care, live negotiation |
| 16–40 | AI supports parts of the task | Client advice, teaching, diagnosis, leadership |
| 41–65 | AI does much of the task with human direction | Analysis, research, design production, coding |
| 66–100 | AI can do most of the task with light review | Data entry, routine writing, transcription, scheduling |
3. Calculate the score
The exposure score is the time-weighted average of the task ratings:
Exposure score = Σ (task share × task rating) ÷ 100
For example, a job that spends 40% of its time on a task rated 80 and 60% on a task rated 10 scores (40×80 + 60×10) ÷ 100 = 38.
4. Separate automation from augmentation
Each task is also labeled “can automate” (AI can increasingly do it end-to-end with light review) or “assists” (AI helps, but a person still does and owns the work). The automation share is the part of the exposure score that comes from “can automate” tasks. This follows the distinction drawn in the International Labour Organization's 2025 research, which found most exposed jobs are more likely to be transformed than fully automated.
5. Add the outlook
Exposure is not the same as demand. Each job also carries an outlook label (growing, stable, shifting or declining) based mainly on the World Economic Forum's Future of Jobs Report 2025 and national labor statistics.
Score bands
- 0–24: low exposure
- 25–44: moderate exposure
- 45–59: high exposure
- 60–100: very high exposure
Limitations
- Scores describe a typical version of a job. Your own mix of tasks may differ; use the custom checker.
- Ratings reflect widely available AI tools, not experimental systems, and they will change as tools improve.
- Exposure does not account for regulation, costs, employer choices or customer preferences, all of which affect how fast AI is adopted.
- A score is not a probability of job loss.
Key sources
- International Labour Organization (2025). Generative AI and Jobs: A Refined Global Index of Occupational Exposure. Working Paper 140.
- International Monetary Fund (2024). AI will transform the global economy and Staff Discussion Note SDN/2024/001.
- World Economic Forum (2025). Future of Jobs Report 2025.
- PwC (2025). Global AI Jobs Barometer.
- US Department of Labor. O*NET OnLine occupational descriptions.
Last reviewed: October 10, 2026. Next scheduled review: April 2027.