Trí tuệ nhân tạo sẽ thúc đẩy năng suất và tạo ra tác động ‘sâu sắc’ đến thị trường việc làm: Bộ Ngân khố

Artificial intelligence (AI) should help Australia achieve its long term productivity target, while having “profound” effects on the labour market.

These are among the conclusions of a detailed analysis of the economic implications of AI prepared by Treasury for Treasurer Jim Chalmers.

As of March, Australia had 162 operational data centres. Another about 130 centres are proposed, taking Australia to nearly 300 operating or planned facilities.

The government is preparing legislation to cover conditions including requirements for the centres on their energy and water use. Data centres will have to bring new renewable energy sources although at last week’s National Cabinet Minister Anthony Albanese agreed to give Queensland and the Northern Territory some flexibility on this.

While the Treasury analysis is generally positive, it is also hedged with uncertainty about the longer term implications of a technology that is still in its early stages.

The analysis says that on current evidence AI will support its assumption of a long term growth in productivity of 1.2% annually. But it won’t be able to do it alone.

In 2024–25 labour productivity declined, and over the past decade it has been mostly flat.

“Reaching the long term productivity growth assumption of 1.2% per annum requires a significant improvement in growth over the transition period,” the paper says.

“The [Productivity Commission’s] quantitative estimates imply that AI will generate some, but not all of the baseline productivity growth in Treasury’s assumption.”

But there are possible risks to this progress, such as trade barriers and geopolitical tensions. Treasury spells out upside and downside scenarios.

There is a plausible upside scenario in which high rates of AI-driven innovation at the global frontier and its diffusion to Australia lift long–term productivity growth to within the range of 1.5–2%. However, there is also a realistic downside scenario in which the pace of AI innovation and diffusion are weak, and offsetting structural headwinds keep productivity growth around its current underlying rate (0.5–0.8%) .

Who’s using AI in Australia?

So far, AI has had limited effects on the labour market, but they will accelerate, albeit unevenly across sectors, the paper says.

“AI will not affect all workers or places equally. It may shift the division of labour across occupations, regions and sectors, depending on exposure, skills and capacity to adapt.”

Information, telecommunications, professional services, finance and insurance are presently having the highest rates of adoption.

While two–thirds of businesses in Australia report adopting AI in some form, less than 10 per cent describe their adoption as significant. This likely reflects the stage of AI development, with many more ways to use AI developing over time.

The paper says AI has the potential to have a bigger impact on high–skilled non-routine and cognitive jobs, augmenting some and automating others. Earlier GPT waves primarily automated physical and routine jobs.

Larger firms may adopt AI faster, while smaller firms may face barriers associated with skills, data, computing capacity, cloud infrastructure and access to foundation models. Because AI has the potential to automate a significant share of cognitive, non-routine tasks currently performed by high–skilled workers, and accelerate productivity and economic growth, labour market impacts could be materially different to past waves of technology. The effects are likely to be profound. The ability of the labour market to adjust will depend on the magnitude and timing of the positive technology shock, how it is adopted and diffused throughout global and domestic economies, and how well policy settings adapt to the changing economic environment.

Treasury says where workers use AI to perform better, output per worker increases. “This supports wage growth, labour’s share of income and employment.”

Implications for interest rates

With AI in a phase of high investment, there could be a change to the “real neutral interest rate”, the paper says. (This rate is the point at which monetary policy is neither restricting or stimulating the economy.)

AI hyperscalers are increasingly resorting to issuing debt to fund their capital expenditure programs, demonstrating that AI-driven demand on the global savings pool is already beginning to grow. This could be partially offset by increased household savings, either as precautionary saving from households who are uncertain about AI’s impact on their future employment and income or in expectation of a longer retirement from an AI advancement driving longer life expectancy.

The pressure would be reduced if productivity impacts are lower than expected or if external factors like geopolitical risks affects investment, the paper says.

Its discussion on interest rates is in a generic context, rather than specifically referring to Australia.

Chalmers said that AI “is shaping up to be the biggest economic transformation in our lifetime.” It would affect every aspect of the economy and society.

“I’m confident we can maximise the economic upside and minimise the risks.”

He said AI would be “front and centre” in the Intergenerational Report to be released in September.

“Almost every advanced economy is grappling with how to lift productivity, but as Treasury puts it, AI is the first substantive and credible productivity growth accelerant in almost two decades.

“Advances in AI are coming thick and fast, with frontier labs reaching new milestones in a matter of weeks, not years.”

Chalmers said the government was acting to ensure Australians reap the productivity, investment and jobs benefits from AI.

“We’re working with the states, private sector and the broader community to build a thriving and sustainable data centre industry, attract investment to train AI models here and make Australians beneficiaries from the entire AI supply chain.

“We’re helping businesses and workers access the tools they need to adopt and adapt to AI, and developing world–leading principles to do all this in the safe, responsible way Australians expect.”

Michelle Grattan does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.