Last week the Australian Bureau of Statistics did something new. It added questions about AI to its quarterly Job Vacancies Survey, and the first answers are out. About 82 per cent of businesses said AI had no effect on their level of job vacancies in August 2026. Four per cent said it had reduced vacancies, and less than one per cent said it had increased them.
After a year of loud predictions about jobs disappearing, that is a quieter picture than most of us expected.
What the numbers actually say
The ABS also asked about the next 12 months. About two thirds of businesses don't expect AI to change their vacancies at all. Around eight per cent expect a decrease and one per cent expect an increase.
The picture isn't even across industries. Information media and telecommunications had the highest share of businesses reporting fewer vacancies because of AI, at 16 per cent. Wholesale trade was next at eight per cent, then manufacturing at six per cent. On the other side, 10 per cent of electricity, gas, water and waste services businesses reported more vacancies because of AI.
The ABS is careful to say these figures cover job vacancies only and don't reflect the effect on existing jobs.
The Reserve Bank reached a similar view on 9 November. Its latest Insight finds little evidence so far that AI has materially disrupted the Australian labour market. It also notes tentative signs of weaker outcomes for younger workers in the most AI-exposed occupations, and it warns that those signs are limited and could reflect other conditions.
Where I'd read it carefully
Both sources are stronger on vacancy and hiring volumes than on how the content of roles is changing. The RBA notes that the share of job ads mentioning AI has risen quickly, although some of that may be changed wording rather than genuinely new requirements. So headcount may be steady while the content of the job moves underneath it.
That matches what I tend to see in the market more broadly: AI shows up as a new expectation inside an existing role before it shows up as a new job title. A marketing manager, a finance lead or a software engineer is asked to work with AI, and the role needs redefining before anyone thinks to hire for it.
If you're hiring into this
Don't treat a quiet vacancy count as a reason to wait. If most employers aren't changing headcount, the competition for people who can genuinely build and run AI is unlikely to ease. My guide to hiring an AI engineer in Australia covers how to approach a market that stays tight.
Rewrite the role before you recruit for it. If AI is changing a job you already have, write down which tasks move, which stay human and who is accountable for the result. AI roles explained sets out the roles in an AI team and the order most businesses add them.
Look at your early-career pipeline. The RBA's note on younger workers is early and tentative, but it's worth asking what entry-level work looks like in your team when AI handles part of it. Training juniors to use AI well is often a better answer than hiring fewer of them.
Match the answer to the problem. Some gaps are best solved by automating, some by upskilling, some by borrowing and some by hiring. The Bot, Build, Borrow, Buy framework is a useful way to test that before a brief goes out.
Where this is going
Now that the ABS has added these questions to its quarterly survey, later releases should give us a clearer trend than a single reading. Until then, the safest assumption is that AI is reshaping what roles require well before it changes how many roles exist.
Recent Articles
Stay updated with our latest articles
