There has been no shortage of bold claims recently about artificial intelligence (AI) and jobs — from mass unemployment to over-hyped distraction. Much of this debate is speculative. Often, coming from the tech giants promoting their own products^[1], it is self-serving.

But beyond the hype, my analysis of new labour market data from the United States shows how AI is already starting to reshape work — and what may soon follow in Australia.

What history tells us about technological change

Exposure to new technology does not necessarily mean jobs will be lost. Technology often reconfigures tasks or boosts productivity.

But when change happens quickly, it can still lead to disruption and widespread job cuts – as seen recently at tech firms Atlassian^[2], Block^[3], and WiseTech^[4].

To make sense of this moment, it is helpful to look back. Researchers have long studied how major “general purpose technologies” reshape economies.

Although each wave of innovation – from steam power to electricity to computing – has its own features, a common pattern emerges. Technological change tends to follow a recognisable trajectory: the technology emerges, adoption spreads, some jobs are displaced, and work is reorganised before a new equilibrium is reached.

The key question is: what stage are we at now?

The emerging jobs data suggest we have already moved into the displacement phase — and are entering a broader reorganisation of work.

What Australia can learn from the US

For Australia, the United States offers a useful leading indicator, because it sits at the frontier of technology adoption. The pattern emerging in the US is not one of widespread collapse, but of uneven and targeted disruption.

Recent data from the US Bureau of Labor Statistics^[5] show the sharpest declines in routine, information-processing roles — especially customer support, administrative work, and software and IT services. There are more moderate declines across marketing, banking, travel and retail.

These are precisely the kinds of sectors that employ large numbers of people in the Australian economy, according to Australian Bureau of Statistics labour force data^[6].

A second group of jobs tells a different story. These roles are not yet shrinking, but growth has stalled.

That slowdown may be the clearest signal of where AI is heading next.

Employment in finance, consulting, management and corporate support has largely stalled after decades of steady growth. These functions underpin modern organisations, suggesting the next phase of disruption may already be taking shape.

The hidden signal: entry-level jobs

Perhaps the clearest warning sign is not layoffs, but a decline in entry-level jobs.

In the United States, the unemployment rate for recent college graduates^[7] has risen to about 5.6% — above the economy-wide unemployment rate (about 4%) and experienced graduates (about 3%). For younger graduates, it is around 7%, with 42.5% underemployed^[8], meaning they are working in jobs that do not require a degree.

These figures demonstrate AI may be taking the jobs of the youngest workers.

The resilience of blue-collar work

AI is also beginning to reshape the composition of the workforce.

For decades, US job growth was led by white-collar work^[9]. But over the past three years, that pattern has shifted: my analysis reveals that blue-collar employment has added roughly one million more jobs than white-collar roles, with manual work rising modestly as office-based employment edges down.

The gains are concentrated in sectors such as construction and maintenance — areas less exposed to current AI capabilities. If sustained, this would mark a significant shift in how work — and opportunity — is distributed across the economy.

Why this time might be different

There are strong reasons to think this transition may be more abrupt than previous eras of technological change.

First, the speed of development is unprecedented. OpenAI’s ChatGPT reached over 100 million users within two months^[10] — one of the fastest adoptions of new technology in history — and its capabilities are improving rapidly.

This compressed timeline leaves far less room for a gradual labour market adjustment.

Second, AI is no longer limited to routine tasks. It is increasingly performing cognitive work once done by professionals — drafting legal documents, writing code, analysing financial reports and generating marketing content. This marks a clear break from earlier technologies, which mainly displaced manual or repetitive work.

Third, AI’s reach is economy-wide. Unlike past technologies that reshaped specific sectors, AI cuts across many industries — from finance and law to logistics and customer service.

What comes next?

The key question is no longer whether AI will change jobs — it already is. The real question is how quickly, and who bears the cost.

As AI anxiety grows^[11], there is an urgent need for a national conversation on policies to navigate the impact on jobs.

This includes the likely need for transitional income support, labour market reskilling at scale, and structural reform of secondary and tertiary education.

The AI job apocalypse^[12] may be overstated. But the early warning signs are already here — and they are increasingly difficult to ignore.

References

1. ^{^} promoting their own products (theconversation.com)
2. ^{^} Atlassian (www.abc.net.au)
3. ^{^} Block (techcrunch.com)
4. ^{^} WiseTech (www.abc.net.au)
5. ^{^} US Bureau of Labor Statistics (www.bls.gov)
6. ^{^} labour force data (www.abs.gov.au)
7. ^{^} college graduates (www.newyorkfed.org)
8. ^{^} underemployed (www.theguardian.com)
9. ^{^} led by white-collar work (www.conference-board.org)
10. ^{^} 100 million users within two months (www.theguardian.com)
11. ^{^} AI anxiety grows (www.nsw.gov.au)
12. ^{^} job apocalypse (www.reddit.com)