Bullock warns AI has yet to lift Australia’s productivity

The productivity promise is still hypothetical

Bullock said the RBA and other central banks had treated AI as a “great white hope” for productivity growth. So far, however, neither the technology nor the investment it has prompted has produced a visible improvement. Spending on data centers and rapid adoption may even be adding to inflation before they add to output.

A study by South Korea’s equivalent of the RBA offers a more modest picture. Employees who started using AI generally produced the same amount of work, but worked 1.5 fewer hours per week.

That is a gain in efficiency for workers, but not yet a gain in measured output. Bullock said productivity could initially fall as people experiment with the technology and work out how to use it. Growth may come later, after companies rebuild their business processes around it.

The government’s intergenerational report, published on Tuesday, assumes that productivity will return to its long-term growth rate of 1.2% a year. Under that scenario, inflation-adjusted economic activity per Australian is projected to rise from $99,200 today to $157,300 in 2066.

Economists consider the assumption unlikely. If productivity grows by only 0.8%, the same measure would reach $136,600 per person in 2066.

Treasurer Jim Chalmers told a Guardian podcast about Australian politics that AI would be the most transformative phenomenon in the lives of the current generation. Bullock said many people consider the government’s forecast unrealistic, but that it should be treated as a target to pursue.

My reading is that the report is using AI less as an observed economic result than as a destination for policy. That can be useful, but it makes the gap between experimentation and actual productivity especially important.

Housing is already showing the cost of growth

Bullock’s comments also cut into the government’s defense of its current migration policy. She said immigration had accounted for almost all of Australia’s economic growth over the past year.

New residents generally work and spend at the same time, Bullock said, so they do not necessarily intensify inflation. Housing is the exception: new immigrants cannot immediately create a matching supply of homes.

One Nation has called for negative net migration for three years and proposed reducing the number of temporary migrants by more than 750,000. The Labor government plans to tighten existing migration rules. Home Affairs Minister Tony Burke described that as “part of the solution to the housing crisis,” while warning that a sharper reduction in migration could “wreck the economy.”

Housing prices fell 3.1% over the past three months. The decline followed three RBA rate increases and a reform of tax concessions for property investors. Bullock also pointed to a sharp fall in housing-loan approvals.

Approvals declined across every borrower category, but the largest drop was among investors. Bullock said that had changed their view of whether housing was worth investing in.

Housing was already “pretty unaffordable” before the decline, Bullock added. She described the current fall as a small retreat after a long period of increases. In her assessment, it sits at the lower end of previous downturns and remains within the range Australia has seen before.

The more interesting question is what happens when the two policy stories collide. Australia is relying on migration for near-term economic activity while expecting AI to deliver long-term productivity, yet one is straining housing supply and the other has not delivered visible output growth. The government’s projections assume the second problem will eventually solve enough of the first.

Bullock declined to say whether she would recommend a rate increase on Tuesday. Traders put the probability of a rise above 90%, from 4.35% to 4.6%. That would take the rate to its highest level in 14 years.