Companion to What AI job loss estimates actually measure
Each relationship explains what connects two topics, which sources support it and where the evidence stops.
Read the full brief →Showing all 5 connections
Lines indicate a relationship discussed in the research. They do not establish causation, strength or probability. Records, our analysis and open questions are labelled separately. Edition 1 · 14 November 2026.
Documented relationship
Exposure research classifies work through occupational and task frameworks.
Evidence limit: A category is not an individual dismissal forecast.
Our analysis
A defensible population estimate requires a documented match between study codes and Census categories.
Evidence limit: Ambiguous mappings must remain visible.
Documented relationship
PUMS documentation supplies variables and weighting guidance for population analysis.
Evidence limit: This brief has not estimated demographic exposure counts.
Documented relationship
The customer-support study evaluates productivity after an AI assistant was introduced.
Evidence limit: Its outcome is not the number of eliminated jobs.
Sources: [4]
Open question
Do changes in output translate into staffing, pay or hours changes, and for whom?
Evidence limit: This relationship must be investigated; it is not established by a single percentage.