Productivity Commission chief Danielle Wood is right to describe artificial intelligence as a “general-purpose technology” in the mold of electricity, the internal combustion engine, or the internet.
Where she errs is in assuming that removing regulatory impediments alone will unlock its benefits for Australia.
History shows that the diffusion of such technologies does not occur automatically; it requires deliberate strategy, patient capital, and public–private collaboration.
General-purpose technologies generate productivity growth not at the moment of invention, but when societies build the complementary assets – skills, infrastructure, institutions, and standards – that enable diffusion.
Innovation doesn’t just happen
Electricity transformed economies only once grids, appliances, and regulatory frameworks were in place. The internet reshaped commerce after decades of government investment in backbone networks and public R&D.
Left to the ‘market,’ these transformations would have been narrower, slower, and more unequal.
Australia today faces the same challenge with AI. Businesses report uncertainty, talent bottlenecks, and limited access to compute as barriers to adoption. Even capable firms hesitate to invest without predictable rules and supportive infrastructure.
Regulatory clarity matters – but by itself, it is insufficient. Without targeted public investment in compute capacity, curated datasets, and workforce reskilling, the productivity gains Ms Wood anticipates will remain elusive.
Australia’s false dichotomy
Ms Wood sets up a misleading choice between ‘getting out of the way’ through deregulation and ‘picking winners’ through industrial policy.
This ignores the global reality: the United States, European Union, China, and even middle-power peers such as Singapore, South Korea, and Saudi Arabia are pouring billions into sovereign AI capabilities, digital infrastructure, and industry-academia partnerships.
These are not vanity projects; they are hedges against dependency, enablers of domestic diffusion, and platforms for global influence in setting standards.
Australia is already an outlier. Its commitments to sovereign AI have been extraordinarily modest. It is inconsistent to champion state support for advanced manufacturing while treating AI as a domain where government has no role beyond “getting out of the way.”
For Australia, the question is not whether government should invest, but how. Limiting the debate to copyright exemptions and planning approvals risks relegating the country to a policy taker, not a rule maker, in the AI economy.
Other nations are actively shaping the rules of the game. If Australia confines itself to deregulation, it cedes influence to others.
Time horizons and leadership
Another weakness in Ms Wood’s argument is her implicit faith in rapid, market-driven diffusion. But productivity gains from general-purpose technologies often take decades to materialise.
Electrification required half a century before its impact was felt in factories; computers took two decades before reshaping office productivity. AI will be no different.
Waiting passively while others set the pace risks leaving Australia a late adopter, locked into foreign platforms and supply chains.
Leadership in AI requires not just reactive regulation, but proactive capability building: sovereign compute access, trusted datasets, translational research pipelines, and targeted support for industries where Australia has comparative strengths – mining, agriculture, healthcare, education, and logistics.
Diffusion of innovation always depends on absorptive capacity, which Australia must build now.
Australia cannot outspend Washington or out-scale Beijing – but nor can it afford to sit back.
Beyond growth accounting
Finally, Ms Wood’s estimate that AI could lift productivity by 4 per cent over the next decade sounds enticing, but it is not destiny. The distribution of these gains depends on choices made now.
Without investment in skills, diffusion will stall. Without attention to sovereignty, dependency will deepen. Without strategic alignment, the promise of AI as a productivity booster could turn into another story of missed opportunity.
Australia currently ranks only mid-pack on global AI competitiveness – 28th of 36 in the Stanford AI Index and 30th in economic competitiveness.
Yet it performs far better on governance, ranking around the top ten per capita for policy and governance measures. This governance advantage is real, but it will mean little unless it is welded to an economic one.
That means plugging gaps in compute and talent, accelerating adoption in trade-exposed industries, and leveraging Australia’s credibility in setting standards across the Indo-Pacific.
Rules travel – but they carry weight only if they are backed by domestic capability and market adoption.
The way forward
Australia does not need to mimic the United States or China. But it must resist the illusion that ‘getting out of the way’ will suffice.
Smart industrial policy – focused, transparent, and evidence-based – is not antithetical to markets; it is what enables them to thrive when technologies are general-purpose, capital-intensive, and geopolitically contested.
If AI is to be Australia’s electricity moment, then policymakers must do more than deregulate.
They must lay the wires, fund the research, and shape the rules that will determine whether the benefits flow widely – or bypass us altogether.
Dr Marina Yue Zhang is an associate professor (technology and innovation) at the Australia-China Relations Institute, University of Technology Sydney.
Do you know more? Contact James Riley via Email.