The Reserve Bank of Australia published a staff research note on 9 November 2026 introducing a way to monitor early AI effects on Australia’s labour market. It finds little evidence of substantial disruption overall, alongside tentative weakness among younger workers in more exposed occupations. For researchers and universities, the contribution is a framework for interpreting emerging evidence without assuming causation.
What the RBA staff note adds
Economist Mike Major’s Monitoring Early Effects of AI on the Labour Market, published as Insight 2026-09, brings together occupation exposure, job advertisements, labour market flows, employment and hours worked. Its two main monitoring channels are occupations whose tasks are more exposed to AI, and young workers and graduates.
The publication is staff analysis. Its disclaimer says the author’s views should not be interpreted as those of the RBA or its Boards. Major describes the exercise as partial and descriptive, rather than a causal assessment of AI’s effects or an assessment of the economy’s full adjustment. The distinction matters because the same occupations can be affected by changing economic conditions and other structural shifts.
How the monitoring works
Major uses Jobs and Skills Australia’s automation exposure mapping to group occupations according to the similarity between their tasks and current AI capabilities. Software developers, clerical workers and administrative workers are among the most exposed. November 2022, when ChatGPT brought generative AI into public use, provides the comparison benchmark in the note’s graphs.
The resulting groups measure potential task exposure. They do not measure whether an employer has adopted AI or changed recruitment because of it. Major also notes that occupations exposed to automation tend to be exposed to augmentation: AI may assist workers as well as perform tasks in their place. A high exposure score therefore does not establish that an occupation will lose employment.
The note’s indicators give a mixed picture. Job advertisements in the most exposed occupations have slowed somewhat more than elsewhere, and employment growth and total hours worked have been a little weaker. However, job-finding rates and transitions from employment into unemployment show little evidence of differences across exposure groups. These measures capture different parts of adjustment, so the evidence cannot be reduced to a single headline about job losses.
Why the younger-worker signal remains tentative
Within the most exposed occupations, younger workers appear to have experienced weaker employment outcomes and larger unemployment increases than older workers. Major does not find the same pattern in less exposed occupations. He also reports that the results are similar when restricting the analysis to market-sector occupations.
There are material reasons for caution. Young workers’ outcomes are volatile and particularly sensitive to the economic cycle. Employment in 2022 benefited from an exceptionally tight labour market, meaning subsequent weakness may partly reflect cooling conditions. Post-pandemic changes in working from home are another possible influence. This is the problem of a confounding variable: a difference associated with AI exposure may also reflect other factors.
Data availability adds a separate constraint. The latest complete occupation data used in the analysis cover February 2026, with partial May data also shown. Major says changes to the Labour Force Survey and the transition from ANZSCO to the OSCA occupation classification mean newer data will need a revised exposure mapping. The November publication date does not mean every underlying series measures November conditions.
Hunter places the findings in a wider research agenda
In her 9 November speech at the UBS Australasia Conference in Sydney, RBA Assistant Governor (Economic) Sarah Hunter said aggregate labour indicators could conceal disruption affecting particular individuals and professions. Liaison contacts nevertheless indicated that AI had not affected most firms’ headcounts and was not expected to do so over the following year.
Hunter described the observed differences between occupations as modest and said it was too early to identify whether AI or other factors explained them. She also stressed that replacing tasks need not replace an entire job. Her speech set out a broader RBA research agenda covering productivity, labour markets, inflation and financial stability.
ABC News reporter Gareth Hutchens reported that four staff papers accompanied Hunter’s speech, covering household wealth, data centre investment, export prices and labour market monitoring. Human Resources Director’s Stephen Owens focused on graduate recruitment, while also reporting the note’s caution about attributing weaker outcomes to AI. These accounts place the junior-worker findings within a wider economic investigation.
What the evidence means for universities and researchers
For university teams interpreting graduate employment patterns, the note’s analytical value is its separation of age, occupation exposure and different labour market measures. As an editorial implication, an institution’s weaker graduate outcomes cannot by themselves establish AI displacement: the national analysis identifies several competing explanations and does not estimate a university-specific effect.
For labour researchers, that distinction connects directly to causal analysis. For research offices assessing claims about AI’s economic impact, the framework also separates capability-based exposure from observed adoption and outcomes. It complements broader contextual resources such as the Stanford AI Index, while answering a narrower question about emerging Australian labour market signals.
What the RBA will monitor next
Major says the RBA will continue developing its monitoring. The note identifies evidence on actual AI adoption, liaison and external business surveys, and wages and earnings as additional ways to understand adjustment. It also points to newer occupation data expected later in 2026. The next analytical step is to connect potential exposure with evidence of firms’ actual behaviour while accounting for other changes in labour demand.
Sources
- Reserve Bank of Australia, Mike Major, Monitoring Early Effects of AI on the Labour Market, 9 November 2026. Checked 2026-11-15.
- Reserve Bank of Australia, Sarah Hunter, What Happens to the Economy as ‘Intelligence’ Gets Cheaper?, 9 November 2026. Checked 2026-11-15.
- Human Resources Director, Stephen Owens, Graduate hiring and AI: RBA spots early warning signs for young workers, 9 November 2026. Checked 2026-11-15.
- ABC News, 9 November business live coverage, including Gareth Hutchens’ reports on the RBA speech and research papers, 9 November 2026. Checked 2026-11-15.
