Could AI Help Australia’s Stagflation Problem, Or Just Continue to Mask It?

Australia’s economy is stuck in stagflation – yes we believe it is; but economists are wondering could AI help Australia and its economy with this? Or will it just mask the problem?

Currently the economy has weak productivity, sticky inflation and a structural slowdown that has pushed the country toward a stagflation‑like posture. You could argue with the data that it is not because the economy is growing, but most of the growth is in AI investment that we are yet to see a financial return from. More on this shortly. But claims are emerging that artificial intelligence could lift the economy’s speed limit and restore the dynamism that has been missing for more than a decade.

The tension between these narratives has sharpened in recent months, partly because the Reserve Bank has begun to treat AI investment as a macroeconomic variable rather than a sector‑specific trend.

Don’t Blame Michelle Bullock, Blame AI For the RBA U Turn

The shift became explicit on June 4, when Governor Michelle Bullock told the Senate that the Bank’s rapid pivot away from further tightening was influenced by the scale of AI‑related investment flowing through the national accounts. She noted that the surge in digital and IT spending had rivalled the peak of the 2012 mining boom, a comparison that would have seemed implausible even a year ago. The implication was clear. AI investment had become large enough to change the Bank’s assessment of the economy’s trajectory.

The national accounts support this view. IT investment nearly doubled in a year. Without that surge, GDP would have been weaker and the quarterly profile would have looked more like contraction than resilience. The question is whether this investment represents a structural lift in the capital stock or a temporary bulge as firms rush to modernise their digital infrastructure.

Answering the question matters because Australia’s long‑run speed limit has drifted lower for more than a decade. The RBA estimates potential growth at roughly 2%. But CBA has argued that widespread AI adoption could lift this to 3%. The difference between 2% and 3% may not sound like much but in the context of the size of Australia’s economy, this is the difference between stagnation and renewal.

We are not seeing the returns from AI (yet)

The challenge is that AI investment does not automatically translate into productivity gains. A recent Bain study found that most companies are missing their AI savings targets because human processes, organisational inertia and capability gaps are slowing implementation. Specifically 40% achieved less than 10% savings – although 14% saved over 20% and 43% just cleared 10%.

Bain’s language was blunt and you didn’t need Bain’s commentary to see the reality: Firms are investing heavily, but the payoffs are elusive. The study noted that companies often underestimate the complexity of integrating AI into workflows, overestimate the readiness of their data systems and struggle to redesign roles and processes around new tools. The result is a widening gap between AI ambition and AI impact.

This finding is highly relevant to Australia. We have a long history of adopting new technologies quickly but extracting productivity gains slowly. The diffusion of computers, the internet and cloud services followed this pattern. Adoption was rapid. Productivity improvement was modest. The gap reflected management capability, workforce training and the structure of the economy. A large share of employment sits in sectors where productivity is difficult to measure or difficult to lift. Health, education, public administration and social services are labour‑intensive and heavily regulated. AI can help, but the gains will be incremental rather than transformative.

AI can mask weakness but may not resolve it

The risk is that AI becomes another episode of capital deepening without productivity improvement. This would lift GDP mechanically through investment spending but leave underlying growth unchanged. The national accounts already hint at this. Output per hour worked has been volatile but broadly flat. Unit labour costs have risen. Real wages have been squeezed. Inflation has been driven by supply constraints and domestic cost pressures rather than excess demand. In this environment, AI investment can mask weakness for a time, but masking is not resolving. We’re not saying AI couldn’t resolve it but masking is not resolving.

There is a second risk. AI could widen the gap between high‑productivity and low‑productivity firms. Australia already has a long tail of small firms with low digital capability. If AI tools are adopted unevenly, the productivity distribution will stretch further. Large firms with capital, data and technical talent will capture the gains. Smaller firms will struggle to keep pace. This would lift aggregate productivity modestly but leave the median firm unchanged. The economy would become more bifurcated. The speed limit would rise for some sectors but not for the economy as a whole.

The optimistic case rests on three mechanisms. First, AI can automate routine cognitive tasks across professional services, finance, logistics and administrative roles. These sectors represent a large share of employment. Even modest efficiency gains can compound. Second, AI can improve decision‑making by reducing information frictions. Better forecasting, better resource allocation and better risk management can lift productivity without large capital outlays. Third, AI can enable new products and services that expand markets. This is the growth‑creation channel rather than the cost‑reduction channel.

The pessimistic case rests on three counter‑arguments. First, AI adoption requires complementary investment in skills, data quality and organisational redesign. These are slow to build. Second, AI tools may deliver private benefits without delivering social benefits. Firms can use AI to optimise pricing, segment customers and extract more value without increasing output. Third, AI may increase demand for high‑skill labour while reducing demand for mid‑skill labour. This can raise inequality and reduce aggregate demand unless wages adjust.

Stagflation makes things tough

Australia’s stagflation‑like environment complicates the picture. Inflation remains above target. Real incomes have fallen. Productivity has stalled. The Reserve Bank has limited room to stimulate. Fiscal policy is constrained by structural deficits and political caution. In this environment, AI is attractive because it promises growth without inflation. Productivity gains allow output to rise without increasing prices. This is the classic supply‑side solution to stagflation.

The challenge is timing. Productivity gains from technology typically take years to materialise. The diffusion curve is slow. The organisational changes required to capture the gains are complex. The risk is that policymakers and markets overestimate the near‑term impact and underestimate the long‑term potential. This creates a mismatch between expectations and reality.

The Bain study reinforces this point. The firms that achieved meaningful AI savings were those that invested in complementary capabilities. They redesigned workflows, retrained staff and improved data governance. They treated AI as a transformation project rather than a technology project. The firms that struggled were those that treated AI as a plug‑and‑play solution. The lesson for Australia is that investment alone is not enough. The country must build the organisational and human capital required to capture the gains.

This brings the debate back to the speed limit. The RBA’s 2% estimate reflects the country’s recent performance. Productivity has been weak. Capital deepening has slowed. Labour supply has grown through migration rather than participation. CBA’s 3% estimate reflects a different view. It assumes that AI adoption will be broad, rapid and effective. It assumes that firms will overcome the organisational barriers identified by Bain. It assumes that the gains will diffuse across sectors rather than concentrate in a few. These are plausible assumptions, but they are not guaranteed.

Could AI help Australia with stagflation? Yes but…

The conclusion is that AI can lift Australia’s speed limit, but only if the country earns the gains. The investment surge is encouraging. The comparison to the 2012 mining boom is striking. The RBA’s recognition of AI as a macroeconomic force is significant. But the Bain findings are a reminder that investment without execution delivers little. Australia’s history suggests that diffusion will be uneven. The country has world‑class firms in mining, finance and logistics. It also has a long tail of small firms with limited digital capability. AI could widen this gap.

Australia’s stagflation‑like environment creates urgency. The country needs a new source of productivity growth. AI can provide it, but only if firms invest in skills, data quality and organisational redesign. The speed limit can rise from 2% to 3%. The gains are available. But they must be earned.