A tool built to catch the first crack
The Reserve Bank of Australia did not set out to declare that AI is or is not taking jobs. In Insight 2026-09, published on 9 November, staff economist Mike Major sets out a monitoring tool: a way to notice unusual movement first in the places AI would be expected to show up, namely occupations whose tasks look most like what current AI can do, and the young workers and graduates who enter through those occupations.
That framing matters, because the youth signal is the part of this note most likely to travel, and it is the part the author handles most cautiously. The paper's own first bullet is the opposite of a jobs warning: there is little evidence to date that AI has materially disrupted the Australian labour market.
What the data actually show
Job advertisements have fallen across every exposure group since November 2022, the month Major uses as a benchmark because ChatGPT brought generative AI into public use. The fall is largest in the most-exposed quintile, and the occupations contributing most to it are general clerks, software and applications programmers, ICT business and systems analysts, call or contact centre workers and receptionists.
The flow measures, which should move earlier than employment totals, do not separate the groups. Major reports little evidence of differences in job-finding rates or in employment-to-unemployment transitions across occupations with different AI exposure. Employment growth and total hours have been a little weaker in more exposed occupations, and unemployment outcomes there are described as more tentative still, because participation decisions sit inside the same numbers.
The youth cut is where a pattern appears. Graphs in the note suggest softer outcomes for young workers relative to the aggregate in recent years, and within the most AI-exposed occupations, younger workers appear to have had weaker employment outcomes and larger increases in unemployment than older workers. In less-exposed occupations, that age split is not present. It is a real pattern in the data. It is also, in the note's own handling, a volatile one.
Three reasons to keep the word tentative
First, exposure is not adoption. The ranking comes from Jobs and Skills Australia's mapping of how closely an occupation's tasks match AI capabilities. It measures potential, not whether a firm bought a tool, changed a hiring plan, or cut a role. The same occupations that score high on automation exposure also tend to score high on augmentation exposure, so a high score on this mapping does not, by itself, point toward job losses.
Second, the timing is awkward. The latest complete occupation data are for February 2026, with partial figures for May. The note itself cautions that youth outcomes are volatile and more cyclical than aggregate ones, and that youth employment in 2022 partly reflected an exceptionally tight labor market at the time. Cyclical normalization is therefore a possible explanation for some of the softening since then. The note raises that explanation; it does not separate it from any AI effect.
Third, the central bank's own business contacts do not report a wave of cuts. In her companion speech on 9 November, Assistant Governor Sarah Hunter said liaison contacts indicate AI has not yet impacted most firms' headcount and is not expected to over the next year, while noting tentative signs of weaker job ad posting and employment outcomes in more exposed occupations, particularly for younger workers.
What a reader should take from it
For a graduate choosing a field, this note is not evidence that software or clerical work is closed. It is evidence of a tentative pattern: outcomes for young workers in exposed occupations have softened relative to older workers in the same occupations, at a time when overall conditions were also easing. This data does not separate those two explanations, and the author does not pretend otherwise.
For employers and policymakers, the useful part is the instrument, not a verdict. Continued monitoring may show whether the tentative split persists, widens or closes; it will not by itself establish the cause. Until then, the honest summary is short: the note reports little evidence of material disruption to date, and it identifies younger workers in highly exposed occupations as the group that warrants continued monitoring.
Sources
- [1] RBA Insight 2026-09, Mike Major, "Monitoring Early Effects of AI on the Labour Market," 9 November 2026Read source
- [2] RBA speech, Sarah Hunter, "What Happens to the Economy as 'Intelligence' Gets Cheaper?", UBS Australasia Conference, Sydney, 9 November 2026Read source