How AI is reshaping jobs and hiring: early signs from Australia

The predictions have been dramatic.

Artificial intelligence will transform productivity. Entire occupations will disappear. Entry-level jobs will become much harder to find.

Some of that may eventually happen.

But the evidence available today is considerably less dramatic.

New research from the Reserve Bank of Australia suggests that AI has not yet caused widespread job losses or produced a major productivity boom across the economy.

There are, however, early signs that the effects are beginning to appear in particular occupations and among younger workers.

The headline employment numbers remain surprisingly stable

AI adoption has moved quickly through businesses.

Employment has moved much less.

In the Reserve Bank's discussions with companies, roughly three quarters said AI had not changed their headcount over the previous year.

Around 12% reported slower hiring or leaving some positions unfilled, while roughly 8% had reduced headcount. A smaller group had actually increased employment.

That is a long way from the widespread displacement sometimes predicted in discussions about generative AI.

The bank's analysis of the early labour market effects, published on 9 November, says there is still little evidence of material disruption at economy-wide level.

The quieter change may be happening in recruitment

Job losses are only one way technology can reshape employment.

Companies can also change their workforce by hiring fewer people.

Instead of replacing an employee with an AI system, a business may decide not to recruit the additional junior analyst, administrator or support worker it once needed.

The Reserve Bank has identified tentative evidence of weaker job advertising in occupations with greater exposure to AI.

The differences remain modest, and researchers stress that other economic factors may be responsible for part of the change.

But it provides a useful warning against looking only at redundancy statistics when trying to understand what AI is doing to work.

Entry-level work deserves particular attention

The clearest early signal appears among younger workers.

Employment outcomes for people aged 15 to 24 have been slightly weaker in occupations with high exposure to artificial intelligence than in less exposed roles.

Again, the evidence is not strong enough to prove that AI is responsible.

But there is a logical reason to watch this group closely.

Entry-level roles often contain a large share of routine tasks.

Preparing first drafts, summarising information, handling straightforward customer enquiries, basic coding and administrative processing are exactly the kinds of activities where generative AI is becoming more capable.

If those tasks take less time, employers may need fewer junior workers even while keeping their experienced staff.

A task can disappear while the job survives

Much of the debate about AI treats employment as a simple choice between a human worker and a machine.

Real workplaces are rarely that simple.

A job is a bundle of different tasks.

AI may automate some of them, make others faster and leave the rest almost untouched.

Jobs and Skills Australia estimates that around 4% of workers are in occupations with high exposure to AI-driven automation.

By comparison, roughly 31% are highly exposed to AI being used to complement or augment their work.

That distinction matters.

A lawyer may spend less time summarising documents but more time interpreting them. A software engineer may write less routine code while reviewing more AI-generated work. A customer service employee may receive suggested responses but remain responsible for difficult conversations.

The productivity miracle is taking its time too

Job losses are not the only prediction that has yet to appear clearly in the data.

The Reserve Bank also finds limited evidence that AI has materially lifted productivity across the economy.

Individual workers can certainly complete some tasks faster, but turning those improvements into economy-wide productivity gains takes more than giving employees access to a chatbot.

Businesses need to redesign processes, change software, train staff and work out which uses actually save time.

Previous general-purpose technologies followed a similar pattern.

Electricity did not transform factories simply because electric motors replaced steam engines. The largest productivity gains arrived when entire production systems were redesigned around the new technology.

AI may follow the same path.

More automation does not mean less work everywhere

There is another side to the transition.

The infrastructure behind artificial intelligence is itself creating demand.

Data centres require construction, power infrastructure, engineering, cooling systems, networking equipment and ongoing technical support.

Companies deploying AI also need people working in cyber security, data governance, product development and implementation.

Some occupations may therefore shrink while others expand.

That makes the employment impact much harder to reduce to a single number.

The biggest disruption may still be ahead

None of this means concerns about automation can be dismissed.

The Reserve Bank itself expects AI to change most jobs in some form and says it is still too early to know the eventual scale of the impact.

The important point is that the transition currently looks more gradual and uneven than many of the most dramatic predictions suggest.

For workers, that may mean adapting to changing tasks before facing disappearing occupations.

For employers, it may mean discovering that the biggest productivity gains require changes to how work is organised rather than simply adding another AI tool.

And for younger people entering the labour market, the challenge may arrive first at the recruitment stage.

The question is no longer only whether AI will replace jobs.

It is increasingly about which parts of those jobs humans will still be expected to do.