Australia should welcome an attempt to get ahead of the AI policy curve. But standards, data centres and new offices are only part of the job. The more urgent test is whether AI makes public services materially better for the people who need them.
I welcome the Albanese Government’s attempt to get ahead of the AI policy curve.
For too long, governments have tended to wait until a technology has created sufficiently obvious harm before responding. There is something refreshing about Australia trying to shape the conditions under which artificial intelligence develops, rather than simply accepting rules written elsewhere.
The Prime Minister’s 15 July speech, AI in Australia’s interests, carries this ambition. It proposes national AI standards, obligations for large data centres around energy and water, stronger protection for creative work, and a new Office of AI within the Department of Prime Minister and Cabinet. Its central proposition is that Australia should shape AI, rather than merely accommodate it. The speech is worth reading in full.
There is a clear desire for Australia to be world-leading on technology policy—perhaps strengthened by its willingness to act early on social-media age restrictions. AI is a much larger, more complicated and more consequential challenge. But the instinct to act early is right.
My hesitation is not that government is moving too quickly. It is whether enough of the public investment will reach the places where it can make the greatest difference.
The risk is not simply over-regulation. Good rules matter: around safety, privacy, copyright, energy, water, accountability and public trust. The greater risk is that Australia develops an AI policy that is highly visible in regulation and infrastructure, but less visible in the everyday systems where Australians actually encounter government.
The more urgent question: what will AI enable government to do better?
The Government’s own National AI Plan points in the right direction. It identifies improved public services as a central objective, including the possibility of faster processing, more personalised support and stronger protections. It explicitly connects AI to closing service gaps in health, disability and aged care. The Government’s response to the Senate AI inquiry makes this commitment clear.
This is where the national conversation needs to become more practical.
In my earlier posts, Fixing My Aged Care's Capacity Blind Spot, and the follow up A Referral Is Not Care: Why My Aged Care Needs to See Capacity and Capability I reflected on how exhausting it can be to navigate systems that are meant to support people at moments of vulnerability. The problem is not that public servants do not care. It is that systems can be fragmented, difficult to understand, repetitive and slow.
AI will not fix poor policy, understaffing or inadequate care. But it could help government do some very human things better:
- explain processes in plain language, tailored to a person’s circumstances;
- prevent people having to repeat their story to multiple agencies;
- identify missing information before an application stalls;
- help frontline workers navigate complex rules and connect people with the right support;
- provide better translation, accessibility and after-hours assistance;
- identify service bottlenecks before they become crises; and
- give citizens a clearer view of what is happening, what comes next and who is responsible.
That is not a futuristic vision. It is a public-service design agenda.
What innovative governments have already shown
Governments do not need to start from scratch. Around the world, some are already using technology to reduce friction at the moments when citizens are most likely to feel overwhelmed: having a child, becoming unwell, caring for someone, losing work or trying to access support.
These are not templates to copy uncritically. Australia has different institutions, laws and communities. But they show a more useful starting point: begin with a human problem, then apply the right technology.
| Government | Human-centred challenge | What technology enabled | Lesson for Australia |
|---|---|---|---|
| Estonia | New parents navigating multiple benefit applications at an already demanding time | When a child is registered, eligible family benefits can be triggered automatically rather than requiring parents to apply separately. Estonia’s proactive-services example | Where government already holds verified information, the citizen should not have to repeatedly prove what the state already knows. |
| Singapore | Families, young people and older people having to navigate many separate agencies | LifeSG brings more than 100 government services into a single life-stage-based platform, including personalised information, eligibility and application-status updates. LifeSG overview | Organise services around life events and needs—not departmental structures. |
| Seoul, South Korea | Older people living alone facing isolation, health risks and the possibility that an emergency may go unnoticed | The city’s smart-welfare programme used opt-in home IoT devices to flag prolonged inactivity or unusual patterns to designated care workers and local support services. In an early rollout, the city reported that alerts enabled earlier intervention in dangerous situations. Seoul’s smart-care case study | The value is not the sensor or algorithm. It is the human response it enables: a timely check-in, social-work intervention, family contact or emergency support. |
| United States | People missing support because they do not know what they may be eligible for | USA.gov’s Benefit Finder asks simple questions and produces a tailored list of possible benefits, including for disability, illness, bereavement, retirement and disaster recovery. Benefit Finder | Technology can help people find the right doorway before asking them to complete a complex process. |
| United Kingdom | Digital services excluding people with low digital confidence, disability, literacy barriers or limited access | The Government Digital Service’s Assisted Digital approach recognises that a digital service must include human support for those unable to use it independently. UK guidance | “Digital first” must never become “digital only”. Human assistance is part of the service, not a costly afterthought. |
| New South Wales | Cities and communities needing better decisions and services in areas such as flood resilience, heat, connectivity and sustainability | NSW’s Smart Places programme combines policy, capability-building, pilots and Smart City Innovation Challenges to test technology and data solutions that improve government decision-making and services. NSW Smart Places initiatives | Australia already has useful local foundations. The next step is to connect smart-place innovation more directly to the difficult human journeys through government services. |
Seoul offers a useful corrective to much of the “smart city” language. A genuinely smart city is not one with the most sensors, dashboards or digital twins. It is one that uses technology to notice when people may be falling through the gaps—and then ensures that a real person responds. Seoul is also not standing still. Its current city roadmap includes Level 4 robotaxis, expanded digital flood alerts and further AI initiatives. The mix is revealing: mobility, resilience and everyday urban services are being treated as connected public challenges, rather than as separate technology demonstrations. A recent overview of Seoul’s second-half 2026 plans is here.
That is the more useful meaning of a smart city. Not technology layered over the city for its own sake, but technology used to make it safer, more responsive and easier to live in.
This is not an argument for surveillance disguised as care. Any comparable Australian initiative would need meaningful consent, strong privacy protections, transparent safeguards and non-digital alternatives.
But the principle is powerful: technology can help make care more attentive and preventative, provided it remains accountable to human judgement and human relationships.
Imagine applying this mindset to aged care in Australia. Not a system that merely collects more data about older people, but one that uses consented information to identify when someone may need support, reduces the burden of navigating fragmented services, and helps carers and frontline workers act earlier.
That is a far more worthwhile ambition than simply adding another AI interface to an already difficult system.
Seoul Wellness robot
Australia has been attempting this for some time
This is not a new idea in Australian public policy.
The Commonwealth’s 2016 Smart Cities Plan argued that Australian cities should be productive, accessible and liveable, with a clear focus on serving citizens. It linked smart technology and real-time data to better services, infrastructure use, energy efficiency and urban outcomes. The Plan is available through the Queensland Parliament’s tabled papers.
NSW’s Smart Places work continues this practical tradition through innovation challenges, pilots, case studies and capability development. It deserves recognition.
The opportunity created by AI is to extend this work beyond infrastructure and urban optimisation into the difficult, often invisible challenges of human navigation: aged care, disability support, health, housing, education, employment and regional access to services.
The question should not be: Where can government deploy AI?
It should be: Where do Australians currently get lost, wait too long, repeat themselves, or give up—and how can technology help restore dignity and agency?
Education matters—but it is not yet the centrepiece
The Prime Minister rightly acknowledged that AI is changing how universities teach and how students learn. He also argued that Australia must invest in people so that AI supports secure and fulfilling work rather than simply replacing it. That is an important direction of travel.
However, the July announcement is not, in itself, a dedicated education or retraining package.
The stronger commitments sit in the earlier National AI Plan: digital and AI skills across the education pipeline; workforce planning; generalist and specialist AI competency units; and engagement between employers, unions and government around workforce transition. The Plan includes universities, schools, TAFEs, community organisations and not-for-profits in its capability agenda.
The most concrete offer remains one million fully subsidised places in a practical, responsible-AI microskills course delivered through the National AI Centre and TAFE NSW. It is a worthwhile start, particularly for workers and small businesses seeking confidence with everyday AI tools. Details are here.
But it is not yet a national programme for redesigning university curricula, supporting academics to adapt assessment and teaching, or giving workers meaningful lifelong-learning entitlements as their roles change.
The risk: funding the intermediaries more than the improvements
There is also a familiar risk in major technology programmes.
Government identifies an urgent challenge, allocates substantial funding, and much of the early spend is absorbed by strategy work, assurance processes, large systems integration and major consulting engagements. Some of that expertise is necessary. Governments must manage risk, procure responsibly and protect the public interest.
But there is a point at which the system begins to fund the machinery around change more generously than the change itself.
The result can be polished frameworks, impressive roadmaps and expensive capability programmes—while the people closest to the problem are left trying to deliver meaningful improvement with what remains.
For AI, that would be a profound mistake.
Australia has capable public servants, universities, social enterprises, small technology firms, community organisations and specialist service-design teams. Many understand particular problems far more deeply than a generalist consultancy ever could: how people navigate aged care, why a student disengages, where a disability-support process breaks down, or why a regional small business cannot take up a new technology.
The Government should retain strategic expertise where it is genuinely needed. But it should structure investment so public value is not diluted before it reaches the frontline.
That means:
- funding practical, small-scale service challenges with clear outcomes;
- reserving meaningful funding for smaller Australian firms, social enterprises, universities and community-led organisations;
- requiring co-design with the people who use and deliver services;
- building capability inside the public service, rather than perpetually renting it;
- creating reusable public infrastructure, standards and tools instead of repeatedly commissioning bespoke reports; and
- publishing what each project costs, who delivers it, and whether it improves the citizen experience.
The test should be simple: after the consultants, procurement processes and governance layers have taken their share, is there still enough money—and enough courage—to make a real difference to the person trying to get help?
If not, then we may be building an AI industry around government, rather than using AI to build better government for Australians.
Lead by making life work better
Australia should set high standards for AI. It should protect creative workers. It should ensure data centres do not externalise their energy and water costs onto communities. It should build sovereign capability.
But we should be wary of treating regulation as the destination.
The real measure of AI policy will not be the number of standards published, offices created or data centres approved. It will be whether a person trying to access aged care, understand a government letter, find a job, navigate disability support or return to study experiences government as more responsive, more comprehensible and more human.
That is where the gold coins should be going.
Australia has an opportunity not merely to regulate AI before everyone else, but to show what it looks like when a government uses AI responsibly to make everyday life work better.