AI is on the agenda in Canberra. In August, the Productivity Commission will release an interim report on harnessing data and digital technology, such as AI, “to boost productivity growth, accelerate innovation, and improve government services.” Shortly afterward, the government will host an Economic Reform Roundtable where AI policy will be up for discussion.
AI developers are aggressively pursuing influence over the new rules. The Chinese government wants to include AI in trade deals. Meanwhile, as the U.S. government seeks to “win the AI race,” U.S.-based tech companies are making their own overtures.
The most ambitious intervention has come from ChatGPT developer OpenAI, which recently hired former Tech Council chief executive Kate Pounder as its local policy liaison. Pounder is also a former business partner of Assistant Minister for the Digital Economy Andrew Charlton.
OpenAI’s AI Economic Blueprint for Australia makes bold projections about the new technology’s impact on the country’s economy, accompanied by a host of policy proposals. However, these claims warrant careful scrutiny, particularly given the company’s clear commercial interests in shaping Australian regulation.
The gap between promise and evidence
OpenAI claims AI could boost Australia’s economy by A$115 billion annually by 2030. It attributes most of this to productivity gains in business, education, and government. However, the supporting evidence is thin.
For instance, the report notes Australian workers have lower productivity than their U.S. counterparts, and then claims (without evidence) this is because Australia has invested less in digital technologies such as AI. However, it ignores numerous other factors affecting productivity, from industrial structure to regulatory environments.
The report also describes supposed AI-driven productivity gains in companies such as Moderna and Canva. However, these narratives lack any data about improved organisational or individual performance.
Perhaps more concerning is the report’s uniformly optimistic tone, which overlooks significant risks. These include organisations struggling with costly AI projects, massive job displacements, worsening labor conditions, and concentrating wealth.
Most problematically, OpenAI’s blueprint assumes AI adoption and its economic benefits will materialize rapidly across the economy. However, evidence suggests a different reality.
Economic impact from AI will unfold gradually
Recent evidence suggests AI’s economic impact may take decades to fully materialize. Studies report that some 40% of U.S. adults use generative AI, yet this translates to less than 5% of work hours and an increase of less than 1% in labor productivity.
AI may not spread much faster than past technologies. The limiting factor will be how quickly individuals, organizations, and institutions can adapt.
Even when AI tools are available, meaningful adoption requires time. People must develop new skills, change the way they work, and integrate the new technologies into complex organisations. The economic impacts of earlier general-purpose technologies, such as computers and the Internet, took decades to fully materialize, and there’s little reason to believe AI will be fundamentally different.
The educational risk
Like Google, OpenAI is also aggressively pushing for AI adoption in education. It has teamed up with edtech companies and launched a new “study mode” in ChatGPT.
The push for AI tutoring and automated educational tools raises profound concerns about human development and learning.
Early evidence suggests that over-reliance on AI tools may condition people to depend on them. When students routinely turn to AI, they risk avoiding the mental effort required to build critical thinking skills, creativity, and independent inquiry. These capacities form the foundation of a thriving democracy and innovative economy.
Students who become accustomed to AI-assisted thinking may struggle to develop intellectual independence. This is needed for innovation, ethical reasoning, and creative problem-solving.
AI applications that help teachers personalise instruction or identify learning gaps may be useful. But systems that substitute for students’ own cognitive effort and development should be avoided.
A multi-partner infrastructure strategy
Australia’s digital strategy will undoubtedly include significant investment in AI infrastructure, such as data centres. One challenge for Australia is to avoid concentrating our investment around a single technology provider. Doing so would be a mistake that could compromise both economic competitiveness and national sovereignty.
Amazon plans to spend $20 billion on local data centres. Microsoft Azure already has significant local capacity, as does the Australian company NextDC. This diversity provides a foundation, but maintaining and expanding it requires deliberate policy choices.
Maintaining multiple data centre suppliers helps keep computing power that is independent of foreign governments or single companies. This approach will give Australia more bargaining power to ensure lower prices, greener power, and local skills quotas.
Diversification provides regulatory leverage as well. Australia can enforce common security standards, knowing no single supplier can threaten an investment strike.
Australia’s AI future
AI technology is developing rapidly, driven by large corporations wielding vast amounts of capital and political influence. It presents real opportunities for economic growth and social benefit that Australia can’t afford to squander.
However, if the government uncritically accepts corporate advocacy, these opportunities may be captured by foreign interests.
Australia’s approach to AI policy should maintain human-centered values alongside technological advancement. This balance requires resisting the siren call of corporate promises.
The decisions made today will shape Australia’s future for decades. These choices should be guided by independent analysis, empirical evidence, and a commitment to outcomes for all Australians.
The Australian government must resist the temptation to let Silicon Valley write our digital future, no matter how persuasive their lobbyists or how impressive their promises. The stakes are simply too high to get this wrong.
Uri Gal, Professor in Business Information Systems, University of Sydney
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Follow us on X, Facebook, or Pinterest
