Research check · 9 November 2026
Australia’s exposed young workers show a warning, not an AI job-loss count
A Reserve Bank of Australia staff analysis finds tentative weakness in job ads, employment and hours in occupations with higher AI-automation exposure, especially for ages 15–24. Its job-finding and employment-to-unemployment series do not show a sustained split by exposure. The analysis is descriptive; it does not observe which employers used AI.
What changed
The Australian comparison adds another country and separates the margins that alarming headlines often combine. In more-exposed occupations, job ads slowed somewhat more; younger workers’ employment and hours look weaker. Yet the flow measures do not show a persistent exposure ranking. All exposure groups gained employment from the November 2022 baseline overall. Complete occupation data end in February 2026; partial data reach May.
A separate U.S. working paper tests five occupation exposure indexes against labor outcomes. Its author-built work-structure index is associated with 0.10 percentage points faster annual growth in occupational unemployment per standard deviation of exposure through September 2026. After sector controls, none of the five indexes shows a statistically clear employment decline in the available occupational data; several show faster median-wage growth.
What these measures cannot prove
Neither study measures actual AI use by the employers in the compared occupations. The Australian youth data are volatile and the 2022 starting point followed an unusually tight labor market. The U.S. index author evaluated their own measure against outcomes already observed when it was built; adding the late-2022 unemployment level weakens its composite result to 0.06 points a year, not statistically distinguishable from zero in that specification. A registered forward test is still pending. An unemployment association does not say whether people entered unemployment, stayed there longer or faced AI-caused displacement.
The current answer
The selected U.S. payroll panel still gives a sharper early-career employment warning. Australian data now show a suggestive but not matching youth pattern, while their job-finding flows and broader Norwegian and U.S. trackers are quieter. These are different populations and outcomes, not votes in one global job-loss tally. Follow the outcome and the comparison before changing a career plan.