When the National AI Centre launched, the pitch was big: accelerate responsible AI adoption across Australian industry. Bridge the gap between research and commercial deployment. Make Australia competitive on the global stage.

That was the plan. Here’s what actually happened.

The Numbers Tell a Mixed Story

In its first full year, the NAIC ran over 40 programs and engaged more than 3,000 organisations. On paper, those figures look solid. Dig into the details and the picture gets more complicated.

Most of those engagements were introductory workshops and webinars. Nothing wrong with awareness-building, but it’s a long way from the deep technical collaboration that was promised. The organisations that genuinely moved their AI capabilities forward tended to be the ones that already had internal data science teams.

The Responsible AI Network grew to over 200 members. That’s a decent community, but membership doesn’t automatically translate into changed practices. I spoke with several members who said the frameworks were helpful for internal conversations but hadn’t fundamentally altered how they approach AI procurement or deployment.

Where It Worked

Credit where it’s due. The sector-specific AI adoption kits have been genuinely useful. The agriculture kit, developed with the Cotton Research and Development Corporation, gave regional producers practical starting points rather than abstract principles.

The partnerships with state innovation hubs in Queensland and Victoria created real pathways for mid-market companies. Several manufacturers I’ve spoken with traced their first serious AI project back to an NAIC introduction.

And the standards work matters. NAIC’s contributions to the Australian AI Ethics Framework and ISO/IEC 42001 alignment give businesses something concrete to build governance processes around.

Where It Fell Short

The gap between ambition and execution shows up most clearly in the SME space. Small businesses were supposed to be a priority, but most NAIC programs assume a level of digital maturity that many SMEs simply don’t have.

There’s also the coordination problem. CSIRO’s Data61 continues to operate semi-independently. The Digital Transformation Agency has its own AI agenda. State governments are running their own programs. NAIC was supposed to coordinate all of this, and frankly, the coordination has been patchy.

Funding is the elephant in the room. The NAIC’s budget is modest compared to equivalent bodies in Singapore, the UK, and Canada. You can’t run a national AI capability program on a shoestring and expect transformational results.

What Needs to Change in Year Two

Three things would make the biggest difference. First, deeper engagement models. Two-hour workshops are fine for awareness, but they don’t build capability. The NAIC needs funded residency programs where specialists work alongside businesses for months, not hours.

Second, proper metrics. Counting engagements isn’t meaningful. Track actual AI deployments. Track revenue impact. Track time from first contact to production system.

Third, coordination with teeth. If NAIC is supposed to be the hub, give it actual authority to coordinate federal and state AI programs. Right now it’s a hub that other bodies can choose to ignore.

The Bigger Picture

Australia’s AI ambitions are real but our institutional capacity to deliver them is still developing. The NAIC is a necessary institution and its first year showed promise in some areas. But promise isn’t enough.

The countries winning the AI adoption race aren’t the ones with the best frameworks. They’re the ones that move fastest from framework to deployment. That’s the gap Australia needs to close, and NAIC needs to be the organisation that closes it.

Year two will tell us whether this was a genuine catalyst for change or just another government initiative that looked better in the press release than in practice.