RBA Governor Says AI Has Yet to Lift Australia’s Productivity

Reserve Bank of Australia Governor Michele Bullock said there is still no evidence that AI is making Australia’s economy more efficient. Instead, she said rapid adoption and data-center investment are so far adding to inflationary pressure, while a possible technology-valuation bubble remains a risk to economic activity.

The remarks put a central banker’s caution beside the large economic promise attached to AI. Bullock said central banks were monitoring whether AI could be a bubble; if technology valuations fell sharply, the resulting hit to wealth and investment could weaken economic activity. She did not say such a downturn was inevitable, only that it was a risk worth watching.

Bullock’s explanation is a sequencing argument. Businesses may spend heavily on AI tools and infrastructure before they have changed the way work is organized around them. During that adjustment, she said, workers and firms can be experimenting rather than producing more. Only after business processes are redesigned might the productivity gains become visible.

[While] people fiddle around and try and figure out what to do with this new technology, productivity actually can decline.

Michele Bullock, Reserve Bank of Australia governor

That near-term drag has a direct inflation channel. Bullock said AI-related investment can add demand before any efficiency benefit appears. In Australia, she has pointed particularly to data-center construction, where investment may compete for already constrained construction resources. In an August press conference, she also said the RBA was monitoring whether AI-related pressure in tight parts of the economy could keep inflation elevated.

• Investment can raise demand for construction capacity and other AI-related inputs before productivity gains appear.
• Experimentation can initially lower measured productivity if firms have not yet reorganized work around AI.
• A sharp fall in technology valuations could become an economy-wide problem rather than only a market event.

Bullock pointed to one indication of what a later payoff could look like. Research from South Korea’s central bank found that employees who adopted AI produced the same output while working 1.5 fewer hours each week. That finding suggests time savings, not necessarily more output: whether those hours become higher production, lower costs, or simply less work depends on what employers do next.

The uncertainty matters for Australia’s long-range economic planning. The government’s intergenerational report assumes annual productivity growth returns to 1.2% and attributes a potential long-term lift to AI. Under its assumptions, inflation-adjusted economic activity per person rises from A$99,200 to A$157,300 by 2066. Bullock’s comments do not reject that outcome; they emphasize that it depends on a productivity transition that has not yet shown up in the economy.