A SIGNAL.
NOT A SENTENCE.
How does an AI exposure score work? Here are the sources and scoring paths behind the Index, what the number can tell you, and where its certainty ends.
Explore 709 occupations ↗First, read the scale correctly.
The replaceability score runs from 0 to 100. Higher values indicate greater estimated exposure of work tasks to automation. It is a directional comparison, not your percentage chance of losing a job.
Your actual task mix, organization, responsibilities, and use of AI can differ substantially from the occupation average.
A title starts the inquiry.
Your tasks add the detail.
The Index uses stored BLS-based occupation data. O*NET, OECD capability indicators, and task-exposure research are contextual references. The derivation of the stored occupation scores is not reproducibly documented in the published dataset.
A personal result can vary by model and between runs. It is not reconciled to the stored occupation baseline. The current generation prompt asks for a provocative tone and urgency; that framing is a product choice, not scientific evidence of imminent job loss.
- Match the work.
The assessment uses your title and description to find a relevant occupation in our BLS-based dataset.
- Estimate task exposure.
When an AI-generated assessment is available, a language model interprets the work you describe. When it is unavailable, an occupation-based fallback uses the stored occupation score and task templates.
- Put the number in context.
The result combines a directional score with task-level explanation and occupational context. Generated recommendations and timelines are estimates to examine, not established forecasts.
Show the source material.
Bureau of Labor Statistics
Occupation definitions, median wages, employment counts, and projections. These describe groups of workers, not an individual employer.
The tasks inside a titleO*NET
Occupational tasks and work activities give structure to the question: what does someone in this role actually do?
A reference for AI capabilitiesOECD capability indicators
The 2025 framework describes nine capability domains across five levels. It provides context; the OECD does not endorse or publish our occupation scores.
Research on task exposureEloundou et al.
The 2023 paper explores occupational exposure to large language models. Exposure is different from actual adoption or job loss.
Useful does not mean certain.
AI capability does not automatically translate into business adoption. Cost, regulation, reliability, accountability, and human preferences all affect what happens in practice.
A historical Index comparison with Frey-Osborne (2013) reported a Spearman correlation of 0.667. The matched records and analysis are not published here for independent reproduction. Agreement with another model is not evidence of predictive accuracy. Read the original paper.
Essays retain their publication and editorial revision dates. The occupation dataset does not provide equivalent source-vintage fields. No score makes a role permanently safe or determines what your employer will do.
The source snapshot does not record the wage-survey year or employment-projection window. Check the original BLS release before treating these figures as current. Editorial changes and export dates do not establish data freshness.
Two numbers. Different questions.
Replaceability (0 to 100) describes estimated task exposure. Lower means less exposure.
Career Health (0 to 10) is a separate ranking aid. Higher means a stronger combination of AI resistance, relative wage, projected growth, and employment scale.
- AI resistance
- 35%
- Relative wage
- 25%
- Projected growth
- 25%
- Employment scale
- 15%
These weights belong to Career Health. They are not the formula for your replaceability score.