AI use and occupationsFatih Kansoy ↗
Occupation observations April–May 2026Five observation windows Definitions

Research & data · 121 countries and areas

AI use and
occupations.

What published AI activity reveals about occupations—and how publication and classification shape the evidence.

May 2026 · median published mass across countries
79.17%occupation massvs42.25%task mass

The occupation facet retains substantially more of the original country usage denominator.

0.996

Rank correlation between published task counts and country usage volume.

01 / Global publication patternInteractive figure

Loading the global publication figure…

What is observed

The source classifies consumer Claude conversations into occupational activities. These are observations of platform activity; users’ jobs and representative workforce adoption are not observed. Measurement definitions.

The research questions

From global patterns
to occupational measurement.

  1. 01

    How does the occupational pattern differ across countries?

    Compare original published shares within a common source classification, with missing observations kept visible.

  2. 02

    What changes when classifications change?

    Inspect the European occupational crosswalk and its alternative descriptions. The mapping work is a regional application of a general measurement problem.

  3. 03

    How do the measures relate to external evidence?

    Examine the US benchmark, occupational activity surveys and theoretical capability comparisons.

Working paper

Publication, classification
and validation.

A study of what can be learned from published occupational activity, with a separate technical construction note.