A lot of the fear – and excitement – around AI is based on an assumption that it will be effective and rapid … like a good painkiller.
People scared of mass job losses, for example, assume AI will – quite soon – be good enough to take over our jobs. And many of those excited about the possibility of kicking back and relaxing (or raking in a lot of profit) by putting AI to work also assume it’s a fast-moving, formidable beast.
Perhaps most promisingly, for anyone who has been listening to economists’ cries about Australian productivity stagnating and prices stubbornly rising, AI is like a knight in shining armour: one that might be able to yank us out of our predicament.
Because here’s the thing: if AI can help us cut costs or slash the time we spend on making things or providing services, it could bring down the costs of production, and therefore the prices we pay for these things. AI could also help slow down price growth by making it easier for businesses to increase supply, bringing it closer in line with demand, and therefore reducing price pressures.
These things would ease our cost-of-living crisis and improve our standard of living. It seems, at least, like AI could be the answer to some of our biggest economic problems.
But like a lot of “silver bullets”, it’s not quite as simple as we might think. And there’s likely to be some economic growing pains before we see any potential benefits.
The Reserve Bank pointed out this week, as it has for many weeks now, that our inflationary woes come from a combination of things (including global oil shocks and conflicts over which we have little to no control) but that, fundamentally, it’s because of an imbalance between supply and demand in the Australian economy (some of which we do have control over).
Some of us might be trimming our spending, especially if we’re getting hit by higher interest payments, but Australian households as a whole – and especially those that are more wealthy – are still spending … quite a lot, including on things that are not strictly necessary.
Meanwhile, because we’ve got a limited amount of resources, and we haven’t improved our ability to get more out of them, our capacity to grow supply has been stuck, stopping us from being able to meet growing demand.
As you know, when demand outstrips supply, prices tend to go up.
Despite nearly one-third of Australian big businesses having adopted AI in 2024-25, up from less than 10 per cent in 2021-22, we haven’t seen noticeable growth in productivity – or job losses.
Of course, correlation (or lack thereof) doesn’t always mean causation. But it’s clear that while job growth has been slowing for some occupations, especially those exposed to AI – such as telemarketers and registry clerks – most jobs remain intact or have just seen some parts of their job requirements shift.
And when it comes to productivity benefits, they remain unremarkable. The uptake of AI hasn’t (yet) led to a miraculous boom in our stagnating productivity growth.
That’s probably for several reasons.
First, the productivity benefits of major technological advancements tend to take a long time to “diffuse” – or spread – across the economy. The capabilities of AI might seem to be growing in leaps and bounds, but like any technology – from computers to the steam engine — getting it adopted and used effectively by a large share of the economy is likely to take more than just a few years.
Despite nearly one-third of Australian big businesses having adopted AI in 2024-25, up from less than 10 per cent in 2021-22, we haven’t seen noticeable growth in productivity – or job losses.
In professions such as investment banking, accounting and corporate development, some firms and employees are already using AI quite effectively to program or code models and sift through large amounts of information, making parts of their job easier.
But in some cases, AI is making people’s jobs harder.
The rise of “AI slop” – low-quality, unoriginal or meaningless content such as reports, research and writing produced using AI – is, in some instances, wasting time and reducing the quality of output.
This can harm productivity and inflation in two ways. First, it can reduce the amount of things we can produce. Why? Because we end up spending more time trying to correct errors, sending work back to people who have used AI, and trying to understand the nonsensical content that has been produced using AI.
The Fair Work Commission, for example, which is responsible for assessing applications made by people who have lost their job and are seeking compensation or their job back, has talked about the huge increase in workload for its staff thanks to a colossal wave of AI-generated applications, many of which are incoherent and end up wasting the commission’s time. This, they’ve pointed out, clogs up the commission’s processes and means people with legitimate claims end up having to wait longer for their outcomes.
Second, AI can worsen our productivity problems because “productivity growth” also accounts for changes in quality. Just because a scriptwriter can now pump out hundreds of AI-generated scripts in a day, it doesn’t mean the quality of the writing will be up to scratch. In fact, the final products may be of substantially poorer quality, meaning the benefits (more scripts) may be outweighed by the costs (terrible writing that audiences don’t want to pay for), leaving us with worse productivity.
The Reserve Bank’s latest forecasts released this week show it expects productivity growth to actually go backwards in Australia for the rest of the year.
The bank didn’t necessarily blame this on AI, but it did note that “stronger-than-expected growth in AI and related technology prices” might flow through to higher costs for companies, which could then be passed on to customers.
And because of the huge amount of energy required to power the data centres used for AI, the bank said it was already seeing some economies facing higher energy costs and consumer prices.
We also have to remember that energy is a resource. If we’re using more of it to produce our goods and services, that means we might actually end up being less productive because we’re using more resources to pump out a given amount of supply.
When asked whether AI was a concern for Australia, RBA governor Michele Bullock said in some sectors such as construction, which are already being squeezed by labour shortages and high costs, the additional demand for building data centres could “keep inflation elevated before we get the productivity improvement that we’re hoping for”.
While Australian business investment – which can help drive innovation and productivity improvements – has grown recently, a large part of this spending has been on data centres and fitting them out.
One of the problems with this is that it requires a lot of construction workers and materials, raising demand for these things and, therefore, their costs (which feed into inflation). And because many of the materials needed to build data centres are imported, these projects are not contributing as much to economic growth as they would if they were using products made in Australia.
Of course, over a longer time period – perhaps a decade or two – we may have figured out the best rules and uses for AI, and it’s very possible that the costs of building and using AI will be lower, leaving us with greater capacity in the economy, higher productivity, and lower prices.
But until we get really good at using AI in our work, understanding the dangers and building guardrails, constructing data centres at a lower cost, and making them guzzle less (or more renewable) energy, the short-term consequence might actually be inferior.
If we don’t manage these things, what we think is a painkiller for our economy might end up being worse – machine learning that exacerbates some of our biggest economic problems.
The Business Briefing newsletter delivers major stories, exclusive coverage and expert opinion. Sign up to get it every weekday morning.
Millie Muroi is the economics writer at The Sydney Morning Herald and The Age covering workplace and economics. She was formerly an economics correspondent based in Canberra’s Press Gallery and the banking writer based in Sydney.Connect via X or email.From our partners
About the Author
