The risk picture: catastrophic if we get it wrong, survivable if we get practical
Lawler’s early modelling, using an occupation-level method, painted a dire scenario: up to 46.5 per cent of local jobs displaced and $4.49bn wiped from gross regional product each year – a Great Depression-scale shock for a regional economy. A more nuanced, task-based lens lowered the displacement to 18 per cent – still a structural adjustment that would bite. “It really does matter because we are looking at very different pictures for our regional economy,” she said. Jobs and Skills Australia’s (JSA) new report, Generative AI and the Workforce, offers a little air: the evidence suggests that entry-level roles remain relatively stable for now, even as AI begins to transform tasks across higher-skill occupations. The findings emphasise that AI’s impact is task-based, not occupation-wide – meaning parts of a job may shift or disappear, but most roles won’t vanish entirely. Still, as Lawler warned, today’s stability doesn’t guarantee tomorrow’s: once adoption matures, the disruption could accelerate.
RDA Mid North Coast CEO & Director Dr Madeleine Lawler.
We’re still at ‘adoption’, not ‘maturity’
The taskforce returned repeatedly to a simple reality: most organisations are still experimenting. There are pilot projects here and there – a chatbot trial in a customer service team, a handful of staff testing ChatGPT, a teacher experimenting with AI lesson prep – but the region is a long way from AI being embedded in workflows or backed by formal governance. Lawler described the journey as a three-step ladder: adoption pilots, where organisations dip a toe in; integration, where AI tools start shaping business processes; and finally, maturity, where usage is systematic, governed and accountable. “We’re barely even at the adoption stage,” she said, warning that the gap between early adopters and laggards could quickly widen into a fault line. JSA’s framework of seven generative-AI “personas” offers a useful way to understand who needs what along that journey:
- Leaders – those tasked with setting direction, guardrails and investment priorities.
- Enabled workers – employees using AI to improve productivity and efficiency.
- Creators – specialists developing new AI applications and content.
- Explorers – innovators testing boundaries and experimenting with emerging tools.
- Educators – teachers and trainers embedding AI into learning and workforce development.
- Supporters – IT, HR and compliance staff who ensure AI adoption is safe and ethical.
- Informed citizens – community members who need enough literacy to make everyday use of AI safely.
Each persona implies different skills pathways: executive training for leaders, micro-credentials for workers, curriculum design for educators, and community literacy programs for citizens. Without targeted investment, the personas at the edges risk being left behind. The point hit home when local stakeholders described their starting points. A TAFE representative admitted it is only just beginning to test AI in course delivery, with individual teachers trialling tools such as Microsoft Copilot but no system-wide policy yet in place. Similarly, council staff acknowledged that while AI could streamline development applications or customer service, they lack both the budget and governance frameworks to proceed safely. The implication was clear: moving from scattered pilots to coordinated maturity will require shared standards, training programs and governance frameworks – things the Mid North Coast currently lacks.
Connectivity is the precondition
NBN Co’s Chris Simon didn’t sugar-coat the starting point. On the Australian Digital Inclusion Index the Mid North Coast scores 68/100 against a national 73 – dragged down by affordability and accessibility – even as data demand has exploded 1100% since 2013. “Broadband connectivity is now as essential as electricity and water,” he said.
There is good news, however, in the upgrades. Around 90 per cent of premises on fibre-to-the-node (FTTN) or fibre-to-the-curb (FTTC) can now upgrade to fibre to the premises (FTTP) – that’s 77,000 more addresses – with wholesale speed profiles stepping up (e.g. 100 Mbps plans lifting to 500 Mbps, 250 to 750) and a headline 2 Gbps option on FTTP. Fixed-wireless capacity has also been lifted, with typical access of 200–250 Mbps on Home Fast and up to 400 Mbps on Home Superfast. For a region defined by its lifestyle appeal, that’s a talent-attraction lever as much as a tech spec.
Telstra: more data, new satellites – and a thicket of red tape
Telstra’s Josh Ford expects mobile data consumption to quadruple by 2030. The 3G shutdown wasn’t a retreat, he said, but a spectrum reshuffle to keep 4G as the workhorse while 5G expands. The company sees low-Earth-orbit satellites (think Starlink) as a complement to terrestrial networks, with “direct-to-mobile” emerging this decade. Telstra covers 2.5 million sq.km, but seamless national coverage still leans on a mix of towers, fibre and space. Then there’s the approvals grind. Ford pointed to the Intercity Fibre Network build: 3,000 land access notices, 1,700 surveys, 1,128 construction certificates and 171 cultural heritage and environment surveys – 6,000 regulatory actions to put 1,000 km of fibre in the ground from Sydney via Canberra to Melbourne. A single mobile tower, even with funding, a willing landowner and materials on hand, still takes a baseline 18 months – most of it in development approvals. “We’ve got to get our act together,” he said, while stressing consumer protections still matter.
Building for equity: UNE’s sovereign platform
If the telcos talked about rails, the University of New England (UNE) talked about trains –specifically, Madgwick, a university-wide platform built to make AI safe, local and useful for staff and students.
“To combat the inequity [and] the shadow AI, we built a sovereign platform… locked down to New South Wales… we can get access to any of the LLMs [large language models] that we want,” said UNE’s Mitchell Smidt. The name is a nod to UNE’s first vice-chancellor, Robert Madgwick, remembered as a curious, forward-thinking leader who embraced experimentation and innovation.
“It’s basically like our little walled garden. We get to do whatever we want inside. Our intellectual property stays in-house… we use our own instances of the LLMs, stored in Australia,” he added. Every user has a private instance, and the models don’t train on interactions. The stack was built with a Newcastle-based company, keeping capability regional.
The motivation is equal parts safety and fairness. With ~16,000 students and 55 per cent studying part-time, UNE couldn’t rely on paid, public tools or tolerate “shadow AI” leaking IP. Smidt’s rule of thumb is pointed: AI must be “safe and reliable… not just a source of confident misinformation.”
A Mid North Coast model?
Smidt’s provocation was bigger than campus IT: “We have the skills, we have the infrastructure, [and] the community of people that can develop an Australian sovereign LLM… built in a regional community.”
That would require power, cooling and water, but it’s not science fiction – it’s a capital-and-planning question. Could the Mid North Coast pilot a region-specific model trained on local industry, health, education and tourism data, under regional governance? RDA MNC’s remit – to convene, de-risk and align – makes it a natural place to start asking.
Missing frameworks, sectoral gaps
By the end of the day, one consensus was clear: there is no overarching framework for AI adoption that links local government, industry, education and health. Each sector is making its own moves, but the lack of coordination risks duplication at best, exclusion at worst. The issue bit hardest for government leaders in the room, who voiced concern that the public sector is falling behind the private in the race to adopt. For councils and agencies already struggling with stretched resources, that gap is a live risk to service delivery.
Guardrails and literacy, not hype
Smidt also warned about AI’s tendency to flatter and fabricate. ‘Hallucinations’ are beguiling precisely because the systems are built to please. The fix is cultural as much as technical: teach people to set clear instructions and guardrails, and be brutally honest about what data leaves the building.
“It’s about building the confidence around necessary adoption… your job isn’t going to be taken by AI, it’s going to be taken by somebody who knows how to use AI.”
What now
Some of the actions are prosaic. Check your bandwidth first – NBN Co’s service checker will tell businesses what’s actually available at an address – and plan upgrades before AI tools arrive. On the labour side, RDA MNC has published an experimental live dashboard. Using JSA data, it tracks shifts across 355 occupations, a signal to watch as adoption moves from pilots to practice. The bigger move is strategic: treat connectivity, skills and governance as one problem – and decide, locally, who the technology is for.
If the day had a thesis, it was this: AI can deepen divides or close them. The difference will be made by unglamorous infrastructure and practical training. And perhaps one big revolutionary idea: a region-first ‘walled garden’ that keeps value – and opportunity – here.
Article: Liz Ginis, Communications & Marketing Manager, RDA Sydney
Here’s what the particpants had to say:
