AI Exposure by SOC Code: Compare the Same Occupation
By Shawn Geeman · 2 min read · Published · Updated
Titles are flexible. Data needs a more stable key. Standard Occupational Classification codes help connect an occupation to its wage, employment, and task context, but a matching code does not make two datasets directly interchangeable.
A code identifies a category, not a person
A SOC code groups occupations by the work performed. A familiar employer title may not match the statistical title. A product manager, for example, can have responsibilities spanning several occupational categories.
Inspect the occupation description rather than choosing the first title that sounds familiar. The rankings link to a page for each listed role, and those pages show the associated code.
Use the code in the public data
Download the CSV or JSON from the dataset guide. The soc field is the code, title is the display name, and replaceability_score is the relative exposure estimate. Keep codes as text so formatting is preserved.
Before joining another dataset, check whether each code appears once, whether broad and detailed occupations are mixed, and whether any records lack a match. Report the number of matched and unmatched records.
Check the classification vintage
Occupational classifications change. A code in one release may have been split, merged, renamed, or assigned a different scope in another. A six-digit match is not enough evidence by itself that the definitions stayed the same.
Use the relevant classification documentation and crosswalk when comparing vintages. The Frey-Osborne comparison essay explains why matching choices matter when comparing automation studies.
Industry averages can hide the interesting part
A major occupational group can include both routine processing and complex responsibility. A group average should not be used as the score for every role in that group.
Also distinguish an average across occupations from an average weighted by employment. One gives each occupation equal weight; the other gives larger occupations more influence. Neither describes every individual worker.
Record the choices that affect the result
A reproducible analysis names the data versions, join key, duplicate policy, missing-data handling, and weighting method. Save your transformation code and the downloaded source files.
Do not interpret a high-exposure category as a forecast of layoffs in a specific company or region. It is a starting point for questions about work, training, and adoption.
Frequently asked questions
Can I search the public dataset by SOC code?
Yes. The CSV and JSON exports include a soc field for each occupation.
Does a code guarantee a perfect match across studies?
No. Classification versions and occupational scope can differ. Check the definitions and any crosswalk.
Sources and methodology
The score is a relative index of task exposure from 0 to 100. It is not a percentage chance of losing a job or a forecast of layoffs. Individual tasks, employer choices, adoption, and regulation all matter.
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.
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.
This essay reflects the evidence available when written; future outcomes remain uncertain.
Discover your free score · Explore the $5 Pro Report · All essays