Job methodology
How AI job-risk scores work
The AI Fear Score for jobs estimates task exposure on a 0-100 scale. Higher means more of the occupation's routine work can plausibly be assisted, compressed or automated by AI.
Score formula
Curated role scores blend two anchors: 55% automation probability (Frey & Osborne, 2013) and 45% LLM task exposure (after Eloundou et al., 2023).
score = 100 * (0.55 * automationProbability + 0.45 * llmTaskExposure)
Long-tail occupations (the O*NET-derived pages) don't have separate published anchors, so both terms use one AI-reviewed exposure estimate — for those pages the score simply equals that estimate, and the page shows a single “AI task exposure” figure rather than two components.
Risk bands
- 75-100: High risk - core tasks are highly automatable.
- 55-74: At risk - a large share of tasks is exposed.
- 35-54: Watch - parts of the role are exposed, but context still matters.
- 0-34: Resilient - hands-on, interpersonal or high-stakes work dominates.
Important interpretation
Exposure is not the same as replacement. A job can have exposed tasks and still grow if AI increases demand, lowers costs or shifts workers toward higher-value work. The score is a way to compare task pressure, not a prediction about any person.