Episode 3: The Punishment of Money

Part One: The World of Work Is Changing Faster Than Recovery

What happens when an injured person is expected to return to a labour market they did not change and cannot control?

In Episode 3 of Shattered, coming later this year, we explore what the rapid transformation of work through AI means for people recovering from workplace injury—and for a workers’ compensation sector racing, like everyone else, to keep pace. Change on this scale will inevitably bring mistakes and unintended consequences. But when decisions affect people’s recovery, livelihoods and futures, what will those mistakes mean for their health and wellbeing?

The premise: Every compensation system is built around an expectation: after injury, people will recover and, wherever possible, return to work. But recovery cannot take place outside time.

An injured person may spend months, or years receiving treatment, undergoing assessments and trying to rebuild their health. While they are recovering, their employer may restructure. Their role may be redesigned. Tasks they once performed may be automated or absorbed into another position. The skills expected of them may change. In some cases, the job itself may disappear.

The worker is recovering. The world of work is not standing still.

This creates a challenge compensation systems have barely begun to confront: what does a safe and durable return to work mean when the labour market is changing faster than the person can recover?

What the evidence does and does not yet show

There is no need to overstate the evidence to recognise the risk. Australia is not yet experiencing economy-wide job destruction demonstrably caused by artificial intelligence. In the year to February 2026, the Australian Bureau of Statistics recorded 257,900 retrenchments. That is a substantial number of disrupted lives, but the annual retrenchment rate was 1.8 per cent—slightly lower than the previous year and far below the levels recorded during earlier economic downturns. The data do not identify AI as the cause of those retrenchments. Australian Bureau of Statistics: Job mobility, February 2026 (Mind you that was February and it is now August).

Jobs and Skills Australia is similarly cautious. Its national analysis found no clear evidence yet that generative AI is causing a broad decline in entry-level roles. Displacement has so far been limited, with most observed effects involving upskilling, redeployment and the evolution of existing work. Jobs and Skills Australia: Labour Market Dynamism

But limited displacement is not the same as limited change. Reserve Bank of Australia research reports that only about 4 per cent of the Australian workforce is highly exposed to AI automation, while a much larger share is exposed to AI-driven augmentation. Nearly 90 per cent of Australian jobs have medium-to-high exposure to augmentation. Firms surveyed by the RBA expected modest near-term reductions in headcount as AI is integrated, alongside substantial changes in roles and skill requirements. Reserve Bank of Australia: Technology Investment and AI

The immediate story, then, may not be the sudden disappearance of work. It may be the quieter restructuring of work: fewer routine tasks, leaner teams, different entry points, new digital expectations and greater value placed on workers who can use AI effectively.

For an injured person returning after a long absence, a job does not need to vanish completely to become inaccessible. It only needs to change enough that their former experience no longer fits.

The entry points may be narrowing

Early evidence overseas gives reason for close attention, although it should not be uncritically transferred to Australia. Research using United States payroll data found no widespread economy-wide displacement from AI. It did, however, identify a widening gap for workers aged 22 to 25 in highly AI-exposed occupations. Their employment stood 19 per cent below where it would have been had it kept pace with less-exposed peers. The difference appeared to arise mainly through reduced hiring rather than increased dismissals. The researchers describe these results as early indicators, not proof of causation. Stanford Digital Economy Lab: Canaries in the Coal Mine?

That matters beyond young workers. Entry-level and transitional roles have often provided a way into a new occupation for people who need to rebuild skills, confidence and work capacity. If some of those pathways contract—or begin to demand capabilities that can only be gained through prior experience—injured workers seeking a new start may face the same closed door from the opposite direction.

They may not be entering the workforce for the first time. But after serious injury, they may be attempting to enter a different part of it.

The race to preserve the original job

Against this uncertainty, the existing employer becomes even more important.

Remaining connected to a workplace can support recovery. Safe Work Australia describes return to work as a complex process that should achieve a timely, safe and durable outcome, with individual, workplace and system factors working together. National Return to Work Strategy 2020–2030

NSW evidence also identifies the worker’s experience with their employer as one of the most significant influences on return-to-work outcomes. Genuine contact, coordinated planning and suitable work can protect the employment relationship while recovery continues. SIRA: Factors influencing return-to-work outcomes

There is therefore a sound reason to help people remain with their original employer wherever this can be achieved safely. But AI and workplace restructuring may create a new pressure: return the person before the relationship, role or opportunity disappears.

That pressure may be felt by the worker, the employer, the insurer and the rehabilitation provider. It may be presented as urgency, early intervention or a concern about long-term employability.

Sometimes it may protect a career. But if urgency overtakes medical readiness, the result may be duties the person cannot sustain, a deterioration in health, a failed return and a damaged employment relationship. The unintended consequences for someone's whole of life is obvious.

A return to work is not successful merely because it occurred. It must also be safe, suitable and durable.

The danger of returning to a job that no longer exists

“Pre-injury duties” can sound like a fixed destination. Increasingly, they may not be.

A professional may return to find that research, drafting or analysis is now completed differently. An administrative worker may discover that routine functions have been automated or centralised. A customer-service role may require oversight of automated systems rather than direct service alone. A manager may return to a smaller team and a broader span of responsibility.

The job title may be unchanged while the job underneath it has moved.

This matters because work capacity is often considered by reference to tasks, hours and restrictions. Yet a worker may be physically or psychologically capable of performing a set of duties while no longer possessing the newly expected mix of digital skills, speed, confidence or recent experience.

That is not a failure of motivation. It is a mismatch created during absence.

Recovery cannot become a race against technology

The policy response cannot be to slow medical recovery to the timetable of organisational change—or to accelerate return to work simply because future options may narrow.

Compensation systems, employers and rehabilitation providers should begin preparing now.

They should:

• assess how the worker’s actual role is changing, not merely whether the old job title remains;

• preserve workplace connection without using that connection to pressure an unsafe return;

• provide AI and digital upskilling during recovery where it is medically appropriate;

• distinguish a temporary skills gap from incapacity or unwillingness;

• examine whether proposed duties are genuinely sustainable in the restructured workplace;

• plan for redeployment before the employment relationship is lost; and

• measure success by safe and durable participation, not the date of first return.

This is urgent because Australia’s return-to-work system is already under strain. The 2025 mid-term review of the National Return to Work Strategy found that return-to-work outcomes did not appear to be improving. It also identified the capacity and skills of the workers’ compensation system as an area requiring greater attention. Safe Work Australia: Mid-term review

AI adds another moving part to a system already struggling to produce consistent outcomes.

A labour market they did not create

It would be wrong to say that AI has already eliminated work on a mass scale. It would be equally wrong to wait for mass displacement before asking what occupational change means for people whose connection to work is already fragile.

Aggregate employment may grow while particular workers lose their place. New jobs may be created while the pathways into them become harder to reach. Productivity may improve while an injured person returns to find that the work they knew—and the career they expected—has changed without them.

The central question is not whether technology is good or bad.

It is whether compensation systems will recognise that recovery now occurs inside a labour market undergoing rapid change and whether injured people will be supported to navigate that change without being blamed for it.

An injured person should not be punished for failing to return to a world of work they did not change and cannot control.

Next in the series: Part Two—When “Suitable Employment” Becomes Theoretical.