Australia cannot afford to panic about AI

The evidence continues to mount that the AI job apocalypse will not arrive. That should be good news for Australia. But after more than a year of dire warnings about mass job losses, and of graduates not being hired, an economic panic has taken on a life of its own.

When Franklin Delano Roosevelt told Americans during the Great Depression that “the only thing we have to fear is fear itself”, he was warning against a feeling of paralysis. As I talk to employers and employees, that is precisely what I worry about, far more than any AI-led job loss. Research from the University of Melbourne last year found Australians ranked the least optimistic towards AI of 47 surveyed countries.

That same research showed Australians held the least belief that AI’s benefits outweigh its risks. This will make the transition to AI more difficult and cumbersome, as low trust in the technology discourages experimentation and open conversation about when and how to use it. A less adaptable workforce will mean slower productivity growth, which Australia cannot afford.

Almost every week we read another news story about a CEO linking job cuts to AI. But we have little evidence about whether those cuts would have happened anyway, even in a world absent of AI.    

Calm must prevail. There is no sign of apocalypse in the data.

Calm must prevail. There is no sign of apocalypse in the data.

A July 2026 report from Australia’s Department of Employment and Workplace Relations finds “no evidence to date of broad labour market upheaval driven by AI in Australia”. The report shows a resilient labour market: even the most AI-exposed occupations have grown 5.6 per cent since late 2022, albeit slower than the 9.5 per cent growth of the least exposed occupations. Internationally, a recent Ramp Economics Lab report shows firms making the largest AI investments increased headcount by about 10 per cent over the two years following adoption. Even OpenAI chief, Sam Altman, reversed his own predictions of doom, saying at a Commonwealth Bank of Australia conference in Sydney that he was “delighted to be wrong” about the scale of white-collar job losses.

This does not mean every job is safe. Roles will change and some work will disappear. But panic does not lead to good commercial decisions. In fact, mistaking disruption for an apocalypse is very risky in its own right.

The first risk is when businesses hear “AI” and think only about cutting costs. Treating AI as a one-way ticket to headcount reduction, rather than as a tool to better equip workers, misses the opportunity for growth and innovation. Last year, the CEO of the Swedish fintech company, Klarna, conceded they “went too far” on AI-driven job cuts, noting that “cost unfortunately seems to have been a too predominant evaluation factor”.

There is emerging international survey evidence to suggest that other companies have similar regrets. A 2025 survey of 1,163 senior business leaders by Orgvue, a workforce planning software company, found that of those who made employees redundant as a result of deploying AI, more than half later admitted they had made the wrong decision.

Second, panic makes businesses impatient. The productivity benefits of AI will take years to show up in the data while companies take time to optimise what they sell and how they deliver it. Economists call this the productivity J-Curve. When a new technology arrives, productivity can look worse before it looks better.The Ramp Economics Lab report brings new evidence of this: hiring gains only emerged six to twelve months after firms adopted AI, once the technology had been absorbed into how they operate.

If businesses demand instant returns from immature tools, they will draw the wrong conclusion about their ability to deliver greater productivity later on. And they will not put in the work required to capture those returns. The dotcom boom showed that investors can overvalue companies while underestimating the long-term effects of a technology. These effects are often bigger, but slower to arrive, than anticipated.

If businesses demand instant returns from immature tools, they will draw the wrong conclusion about their ability to deliver greater productivity later on.

Third, panic makes people unproductive. Fear changes how people work, making them more cautious and defensive. Recently, Harvard published research showing that some avoid AI because it threatens their expertise and status. Others will simply choose to switch off.

In many organisations, employees already do two jobs. The first is the one they are paid for, and the second is the job of self-preservation: managing impressions, avoiding blame, hiding mistakes and navigating politics. AI panic makes that second job bigger.

We do not usually think about the day-to-day minutiae of work as a national productivity question. But we should. A group of academic economists studying almost two decades of findings from the World Management Survey, overseen by the London School of Economics, finds up to a third of cross-country and within-country productivity gaps can be attributed to management practices.

This becomes even more important in the transition to widespread use of AI. Firms should not let fear dictate their future. A panicked rush to conclusions, and a workforce that absorbs that fear by osmosis, will be among the most significant bottlenecks to successful AI adoption.

History overwhelmingly demonstrates our ingenuity to adapt and to imagine a more productive world.

History overwhelmingly demonstrates our ingenuity to adapt and to imagine a more productive world. Leaders must focus on growth before cost. They must put in the work to adapt their organisations before expecting returns. And they must share that optimism with their employees.

There is no bogeyman coming to take our jobs. But fear of job loss could rob Australia of the productivity gains it so desperately needs.