If you graduated this year and are chasing your first role in administration, software development or a contact centre, the job market may feel harder than the headline unemployment figures suggest. Reserve Bank of Australia (RBA) research now has early data pointing in the same direction, and it is specific enough to be useful for both graduates and the HR teams deciding how many entry-level positions to open.

In an RBA Insight paper, RBA economist Mike Major set out a new framework for tracking AI's early effects on Australian labour markets. He compared labour-market outcomes before and after November 2022, when ChatGPT launched. Within the most AI-exposed occupations, younger workers recorded weaker employment outcomes and larger rises in unemployment than older colleagues in the same kinds of roles. The pattern held even as the aggregate numbers stayed calm.

This is not a mass-redundancy story. It is a story about who secures the first step in a career, and that matters both to anyone starting out and to every employer that will need experienced mid-level talent five years from now.

What the RBA actually found

Rather than waiting for the overall unemployment rate to shift, the RBA is monitoring two groups where AI-driven change is expected to register first: occupations whose task profiles most closely match what current AI tools can handle, and young workers and recent graduates.

At the aggregate level, the picture looks stable. Job-finding rates and flows into unemployment appear similar whether or not an occupation is heavily exposed to AI. The divergence emerges when outcomes are split by age: in the most exposed roles, younger workers have fared worse, a pattern absent in less exposed occupations.

Job advertising data reinforces the same reading. Among the biggest contributors to the fall in new job ads tracked by Jobs and Skills Australia (JSA) are general clerks, software programmers, ICT analysts, contact centre workers and receptionists. JSA estimates that about 4% of Australian workers are highly exposed to AI automation, compared with 31% who are highly exposed to AI augmentation, where technology is more likely to reshape tasks than take over the role entirely.

Speaking at the UBS Australasia Conference in Sydney, Sarah Hunter, the RBA's assistant governor (economic), drew a line between the two effects:

"Even where AI replaces tasks, it may not replace the whole job."

Two caveats keep the findings provisional. The latest complete occupation data only run to February 2026, and the Australian Bureau of Statistics (ABS) is migrating to a new occupation classification system, OSCA, which means the tracking methodology will require revised mapping before longer run-comparisons are reliable.

Employers are trimming the intake, not the workforce

The RBA's conversations with businesses through its liaison programme reveal how quiet the adjustment has been. Roughly three-quarters of liaison firms reported no change in headcount attributable to AI over the past year. About 12% were hiring more slowly or leaving roles unfilled, around 8% had reduced headcount and about 6% had added staff.

That 12% figure is the one worth pausing on. Slower hiring and unfilled roles never register as layoffs in the official data, yet they close precisely the doors that graduates and career starters walk through. A July report on AI and employment from the Department of Employment and Workplace Relations (DEWR) reached a compatible conclusion: no broad AI-driven upheaval, but measurably slower growth in occupations exposed to automation.

The broader macro picture adds noise. ABS Labour Force Survey data show unemployment rose to 4.7% in September, the highest reading since October 2021, while the trend measure held at 4.6%. Youth unemployment climbed 0.3 percentage points to 11.1%. On 3 November the RBA raised the cash rate to 4.85%, its fifth increase this year, on a six-to-three vote. With borrowing costs at that level, it becomes genuinely difficult to separate cyclical caution from structural change driven by AI, and that ambiguity sits at the centre of every workforce planning conversation right now.

Signals from individual employers are uneven. Canva announced that 1,271 roles, or 22% of its workforce, would be cut, changed or relocated, and co-founder Melanie Perkins told staff that incremental adjustments would not be sufficient. A CFO survey by the Oliver Wyman Forum and the NYSE found that 70% of CFOs plan to increase their focus on succession planning, while 64% expect to shift away from junior roles. Yet the Australian HR Institute's (AHRI) Australian Work Outlook found that 41% of organisations said AI had increased their entry-level roles, against 19% reporting a fall. The picture is split, and that split is itself informative: the effect depends heavily on how individual employers choose to redesign work rather than on some uniform technological imperative.

For employees and graduates: what this means for you

If you are early in your career and working in, or aiming for, one of these occupations, the data does not say your job is disappearing. It says the tasks inside it are shifting, and that some employers are more cautious about bringing in new juniors than they were two years ago. That calls for a sharper search strategy rather than alarm.

• Lead with what AI cannot own end to end. Dealing with clients, exercising judgement on complex or ambiguous cases, coordinating people across a project and resolving problems that fall outside any template: these are the parts of clerical, coding and contact-centre work that sit firmly on the augmentation side of JSA's framework.
• Show you already work with AI tools. In applications and in every interview, offer a concrete example of how you used an AI tool to work faster, check your output or flag an error, rather than simply listing it as a skill on a resume.
• Ask about the junior pipeline. Before accepting an offer, find out how many entry-level people the team hired in the past year, what structured training they receive and where they typically sit after 24 months. Vague answers are informative in themselves.
• Read employee reviews from people in junior roles. Verified reviews from people who have held those roles tell you whether a company still invests in newcomers or expects them to arrive fully formed.
• Widen your search to adjacent roles. Roles that draw on the same underlying skills but sit in teams with growing remits often represent a better entry point than the job title you originally trained for.

And if rejections accumulate, keep the structural context in mind. With youth unemployment at 11.1%, a slow search reflects current market conditions as much as it reflects your individual candidacy.

For HR: why under-hiring juniors now can hollow out your pipeline

The CFO survey data captures an internal contradiction that HR leaders are well placed to surface to the board. Most finance leaders want stronger succession planning, and most also expect to rely less on junior roles. Those two objectives pull directly against each other: the managers and specialists a business will need in five years are the graduates and trainees it hires, or passes over, right now.

AHRI chief executive Sarah McCann-Bartlett noted that HR professionals would recognise the growing importance of investing in career pathways for roles in AI-exposed areas. In practice, that recognition needs to translate into protecting junior hiring as a long-horizon capability decision rather than treating it as a discretionary line to trim when budgets tighten.

The Fair Work Commission (FWC) reviews Modern Award minimum conditions, including apprenticeship and traineeship classifications, on an ongoing basis. Organisations that have redesigned junior roles in response to AI-driven task changes may want to review whether their classification structures still reflect the work being performed, particularly under the Clerks Private Sector Award and the Information Industry Award, which cover many of the occupations the RBA has flagged.

• Separate cyclical from structural. Audit every frozen or unfilled entry-level requisition from the past 12 months and document the reason: budget pressure, reduced workload, or tasks now handled by AI tools. Only the last category is structural. The others may resolve as the rate cycle turns.
• Redesign junior roles rather than deleting them. If AI now handles the routine component of a clerk or analyst role, rebuild the position around review, client contact and problem-solving. Newcomers still need to learn the business, and that learning requires structured exposure.
• Set a minimum intake and make the cost of skipping it visible. Commit to a floor for graduate, apprenticeship and entry-level hiring that holds across the rate cycle. Model what the talent gap costs in three to five years and put that figure in the workforce plan.
• Publish progression pathways explicitly. Make clear what the trajectory from a junior role looks like at 12, 24 and 36 months. The AHRI Australian Work Outlook data suggest this is a retention lever as well as a recruitment signal.
• Apply the RBA's age-split logic internally. Compare hiring rates and attrition for younger and older staff in your most AI-exposed roles every quarter. If a gap is opening, it is better to see it in an internal dashboard than in an exit interview backlog.

What to watch next

The next data points are close. October labour force figures from the ABS are due on 19 November, and the RBA's next rate decision is scheduled for 8 December. Watch the youth unemployment detail and the occupation-level breakdown ahead of the headline rate, because that is where the RBA's own framework expects AI's effects to register first.

For graduates and job seekers, the practical takeaway is to target roles where AI augments rather than replaces, and to probe prospective employers about how they develop juniors before accepting an offer. For HR teams, the priority is to protect the entry-level intake before a year of cautious hiring turns into a structural gap in the future management bench. The RBA has flagged the early signal. What organisations do with it over the next 12 months will shape their capability position well into the 2030s.