The New South Wales Government published its updated AI strategy last month, and I’ve spent the past few weeks going through it in detail, talking to people inside the public service, and comparing it to what other Australian states are doing.

The short version: it’s ambitious, largely sensible, and almost certainly underfunded for what it’s trying to achieve.

What the Strategy Promises

The strategy outlines four priority areas: government service delivery, economic development, research and education, and responsible AI governance.

On service delivery, NSW wants to use AI to improve outcomes across health, transport, education, and customer-facing government services. The specific targets include reducing average wait times for government services, improving predictive capabilities for infrastructure maintenance, and personalising interactions with Service NSW.

On economic development, the strategy aims to position NSW as Australia’s AI hub. This includes attracting international AI companies, supporting local AI startups, and building an AI-skilled workforce. There’s a proposed AI innovation precinct in Western Sydney, co-located with the new airport development.

On research, the strategy emphasises partnerships between NSW universities and industry, with funding for applied AI research in priority sectors. Health, climate adaptation, and urban planning are the focus areas.

On governance, NSW plans to establish a state AI advisory council, develop mandatory AI risk assessment processes for government deployments, and publish an AI transparency register listing all AI systems used in government decision-making.

What Works

The governance elements are genuinely strong. The proposed AI transparency register, if implemented properly, would make NSW the first Australian state to publicly disclose its AI deployments. That’s a meaningful commitment to transparency that the federal government should match.

The focus on applied research rather than pure research is pragmatic. NSW doesn’t need to compete with Stanford on fundamental AI research. It needs AI solutions to specific problems: managing growing urban populations, adapting infrastructure to climate change, and improving health outcomes for diverse communities.

The workforce development plans include practical elements beyond traditional education. Micro-credentials, industry partnerships, and retraining programs for workers displaced by automation. These are the kinds of programs that actually build capabilities rather than just awarding certificates.

And the explicit inclusion of regional NSW is important. AI benefits that concentrate in Sydney don’t help a state government that represents communities from Broken Hill to Byron Bay. The strategy includes specific provisions for AI adoption in regional industries and government services.

What Concerns Me

The funding doesn’t match the ambition. The announced investment is significant by state government standards but modest compared to what the strategy aims to achieve. Building an AI innovation precinct, funding applied research, establishing governance infrastructure, and retraining workers simultaneously requires substantially more than what’s been committed.

The implementation timeline is aggressive. Establishing an AI advisory council, developing mandatory risk assessment processes, and building a transparency register are each twelve-month projects on their own. The strategy puts all of them in the first implementation phase alongside several other initiatives.

Inter-agency coordination remains the perennial challenge. NSW government AI initiatives span multiple departments and agencies, each with their own IT systems, procurement processes, and priorities. The strategy acknowledges this but doesn’t clearly resolve how coordination will work in practice.

And the economic development aspirations compete with Victoria, which has been actively courting AI companies for years. Melbourne has an established AI ecosystem around the University of Melbourne, Monash, and a cluster of AI companies in the inner suburbs. NSW’s proposition needs to be differentiated, not just another innovation precinct announcement.

Comparison With Other States

Victoria has been the quietest achiever in state AI strategy. While making fewer announcements, they’ve built genuine AI capabilities in health (the Victorian AI health trials) and public safety (emergency services prediction and resource allocation).

Queensland is focused on AI in resources, agriculture, and environmental management, which plays to the state’s economic strengths. Their AI strategy is narrower but more focused than NSW’s attempt to cover everything.

South Australia has partnered with the Australian Institute of Machine Learning (AIML) at the University of Adelaide, creating a research-industry nexus that punches above the state’s weight. They’ve attracted international AI companies with a combination of research talent and lower operating costs than Sydney or Melbourne.

Western Australia is in early stages but has obvious opportunities in mining AI, where the state’s industry leadership creates natural demand.

What NSW Should Do Differently

Three recommendations.

Focus. The strategy tries to do too many things. Pick two or three areas where NSW has genuine competitive advantages and invest deeply. Health AI and urban infrastructure AI are obvious candidates given Sydney’s scale and the state’s hospital network.

Fund properly. An underfunded strategy is worse than no strategy because it creates expectations it can’t meet. Either increase the funding to match the ambition or reduce the ambition to match the funding.

Measure outcomes, not activities. The strategy’s KPIs are heavy on activities (number of programs launched, number of partnerships established) and light on outcomes (measurable improvements in services, actual AI deployments in production). Activity metrics let you look busy without delivering value.

The Bottom Line

NSW’s AI strategy contains good ideas and genuine commitment to responsible AI governance. The transparency register alone would be a significant achievement. But translating strategy into reality requires execution discipline, adequate funding, and honest measurement.

I’ll be watching the implementation closely. The strategy gets a cautious pass. The execution will determine whether it earns a passing grade.