Is AI Reducing Job Vacancies In Australia? Businesses Expect More Than Twice What They Can Measure

Mostly no, and there is finally a number attached. Is AI reducing job vacancies in Australia? On the Australian Bureau of Statistics’ first measurement, published this morning, 3.8 per cent of businesses said artificial intelligence had lowered their vacancies. Another 81.8 per cent said it had changed nothing at all.

That is a much smaller number than the one most founders are quietly planning around.

Outstaffer employs people where they live, so what a role costs to carry is what it sells. Worth saying before the arithmetic rather than after it.

Nobody had asked Australian businesses this before

The Job Vacancies release for August 2026 is the first to carry estimates of AI’s effect on hiring demand. Years into the argument, this is reading number one.

The split: 81.8 per cent no impact, 3.8 per cent a decrease, 0.6 per cent an increase, 13.0 per cent unsure.

What does the unsure number actually say?

That more businesses cannot answer it than answered yes, by nearly three and a half to one. Most reported no impact at all. The figure underneath is the 13 per cent who could not say either way.

The forecast runs at more than double the measurement

Businesses were also asked what they expect over the next twelve months. Across all industries 8.2 per cent expect AI to reduce their vacancies, against the 3.8 per cent who say it already has.

In professional, scientific and technical services the gap is wider: 5.2 per cent measured, 19.7 per cent expected, almost four times. Information media and telecommunications reported 16.2 per cent and expects 43.1. Financial and insurance services, 4.7 against 18.7.

Does any industry expect AI to add roles?

Professional services does, at 7.4 per cent, more than three times the next highest industry. The same industry carries the second largest expected decrease. Most of it named no direction either way.

Your vacancy line already moved, and this data cannot tell you why

This is the part that should bother anyone doing headcount planning this quarter.

Vacancies in professional, scientific and technical services, unadjusted, fell from 35,700 in May 2026 to 29,500 in August, 6,200 gone in a single quarter, a fall of 17.4 per cent. Against May 2025 the fall is 8,900 positions, or 23.2 per cent. Nationally there were 325,000 vacancies in August, down 0.9 per cent on the quarter and 31.2 per cent below the May 2022 peak.

Now try to attribute any of it. You cannot. The 5.2 per cent counts businesses, not positions, so it will not subtract from 6,200, and there is no earlier AI reading to compare against. This release settles it neither way. Nationally, vacancies sit 31.2 per cent below the May 2022 peak and are no lower today than in February 2025.

Which is where adoption was almost universal and depth almost nowhere left things nearly three weeks ago, and the same shape as how far industries diverged on output per hour.

The strongest case for planning on the expectation anyway

It is a decent case. These forecasts come from the people who make the decisions, so 19.7 per cent is partly a statement of intent, and intent moves hiring whether or not the technology delivers. The case also rests on lag: a tool bought in March would show up in a role not posted in December.

Three caveats sit under it. One reading, no trend. The proportions exclude businesses with fewer than five employees and ignore non-respondents. And a survey answer about attribution is a judgement, not a measurement.

Related reading

• Which Media Agency Tasks Can Be Automated
• What Should A Small Business Automate First

Freeze the role or fill it, but price the guess

Holding a role open because AI might remove it next year means paying the cost of the vacancy now against a reduction only 8.2 per cent of businesses expect. The question is not whether AI is reducing job vacancies in Australia. It is what your wrong answer costs, and a role carried at a lower cost base is a cheaper wrong answer.

So test it on a real role rather than a forecast. Run one through and see what it costs, or hold the seat empty another quarter, which is also a decision.

Which roles have you not filled this year, and can you honestly say AI is the reason?