AI at Work in Australia: The Need for a National AI Summit
An OCNUS Consulting White Paper
Executive Summary
In the early 1980s, Australia was in economic decline. Commentators warned that we could end up like Argentina, a once wealthy nation languishing in economic and social malaise. The Hawke–Keating governments achieved a turnaround by fundamentally restructuring the economy. This was only achieved through creating a consensus that brought together government, business, unions, institutions and community groups.
The AI revolution poses different challenges to the economic circumstances of the 1980s, but it demands a similarly nationwide approach.
Artificial Intelligence is reshaping every sector from mining to healthcare. Credible forecasts indicate that generative AI could add tens of billions of dollars to national output, but this benefit is far from assured. Skills are scarce, infrastructure is constrained, and AI uptake varies considerably between regions and industries. Workers are anxious, research is underfunded, and government policy is fragmented and lacking coherence.
Without a national plan, gains from AI will pool in a handful of large firms and capital-city postcodes. Disruption in the workforce could lead to industrial strife and social disharmony. Disparities in access to AI will magnify existing inequalities.
Australia, therefore, needs a National AI Summit to develop a strategy akin to the Hawke–Keating consensus. It would set shared goals, lock in investment for training and energy upgrades, spread adoption and give workers a say in how technology is used. Only a pact of this breadth can turn AI’s promise into broad-based and long-term prosperity.
Economic Impact: Potential and Risk
Optimistic forecasts abound. Microsoft and the Tech Council's scenarios predict an AI-driven GDP uplift of $ 45-115 billion, with professional and financial services alone capturing up to $ 13 billion and retail up to $ 9 billion.
Evidence on the ground is catching up. A 2024 PwC study found that sectors with high AI exposure are already posting labour productivity growth almost five times the economy-wide average. Yet Australia still pays a modest six per cent wage premium for AI skills, eight points below the global benchmark.
Businesses that delay AI integration will quickly lose to their competitors who do. The winners will be organisations that invest in people and embed responsible practice early. Boards must treat advanced digital training and recruitment as strategic priorities.
While waiting for the government to catch up on regulation, organisations should act as if mandatory rules will land within two years because retrofitting compliance is costlier than building it in from day one.
Corporations should tie their AI plans to their energy strategy. New AI workloads could arrive just when the grid is under maximum strain. AI power resilience needs to be added to 2025 risk registers.
Large corporations would do well to engage with network providers, consider locating facilities near renewable energy zones and invest early in efficiencies to cut power costs and reduce risk.
AI Adoption: Different Speeds and Inequality
The headline numbers mask a divided market. Fifty-two per cent of firms now use at least one AI tool, but adoption among small and medium-sized enterprises sits at only 40 per cent.
Large enterprises dominate implementation, whilst many smaller firms remain stuck at the pilot stage. Services companies lead with 56 per cent adoption, whereas industrial businesses sit at 38 per cent.
Policy incentives can make a difference. Targeted grants helped agriculture lift adoption by twenty-one points in the final quarter of 2024, yet manufacturing slid from 45 per cent to 30 per cent over the same period despite similar attention. Meanwhile, there is a wide disparity between uptake in cities compared to regional areas.
If Australia continues with the current pattern of AI adoption, the benefits will concentrate in narrow sectors, large firms and cities. This will compound existing inequalities with their attendant economic and social consequences.
The Workforce: Transformation Not Extinction
Mass technological unemployment is forecast by some, but it is not apparent in present trends. The evidence points to task reshaping rather than wholesale job loss. Current tools can potentially automate or augment 44 per cent of task hours. In technical but routine office work, this rises to 62 per cent today and could reach 98 per cent by 2030.
ServiceNow estimates that 1.3 million roles might be replaced by 2035, yet 885,000 roles will be deeply enhanced, and a further 369,000 new positions will be needed to run the technology.
Significant disruption in white-collar work is likely in the medium term. Blue-collar work will also be affected more slowly, but will occur when robotics catches up to AI. Optimists like to quote the Jevons paradox, but rather than relying on that, a more prudent approach is to plan deliberately for the impact of AI at work.
Modern economies have demonstrated high degrees of flexibility if skilfully managed. The national imperative is to reskill at speed, blending data fluency with domain expertise and ethics so people and machines can work side by side.
Trade unions: Including Workers in the AI Consensus
The ACTU models that 7.2 million employees – about half the workforce – will need to reskill for generative AI, with 3.3 million jobs augmented and 3.9 million disrupted. To avoid future industrial strife, trade unions will need to be positively included in the policy development for the AI transition.
Three initiatives will help enlist unions in a nationwide AI strategy. First, bargaining rights on technology should be guaranteed, so unions can negotiate over the introduction, data use and related practices of AI systems. Second, there should be established a tripartite “AI Skills Transition Fund” to deliver short courses designed by unions, TAFEs and industry. Third, employers should be mandated to provide plain-language impact statements and safety assessments before deploying AI.
Education and Training: Building AI Capability from Class to Campus to Work
Evidence from schools and universities shows a cautious but steady AI uptake. Australia leads some of our peers in the Anglosphere. Classroom adoption here is at 78 % for secondary schools, ahead of 60% in the United States and 50% in England.
Universities report that 75 per cent of academic staff now use generative AI for teaching, research or administration. Tertiary students are even more enthusiastically adopting the technology.
However, there is no reason for complacency. The International Data Corporation puts generative-AI adoption across educational organisations at 86 per cent globally, which is the highest of any industry. Australia’s relative performance is good, but not at the leading edge.
Much of the AI use in teaching is only at the pilot level. Little has been developed and implemented in delivery, feedback and assessment. Many educators are suspicious of AI, and there’s no doubt the consequences for learning are significant. However, AI is not going away, so the technology needs to be harnessed and controlled, not denied and resisted.
Another matter that needs addressing is that AI penetration in education is highly uneven, with significant disparities based on regionality and socioeconomic demographics.
Professional development in the workforce is lagging. A KPMG survey shows only 24 per cent of Australians have taken any AI-related training, well below the global average of 39 per cent.
Upskilling the population at all levels is imperative. Increasing teacher capability in all sectors needs to be prioritised. Curricula, delivery and assessment have to be quickly reformed to accommodate the new reality. The access gap should be addressed so that AI education is not limited by socioeconomic and regional factors. AI training in industry needs to be rapidly accelerated.
Energy and Infrastructure: The Hidden Constraint
The International Energy Agency expects global data centre demand to exceed 945 TWh by 2030, more than Japan uses today. However, meaningful modelling in Australia of the impending extra electricity demands has not yet been done.
A radical restructuring of electricity generation is necessary, and it needs to happen soon. Australia’s AI ambitions cannot ride on an unstable electricity grid. Investment in renewables, storage and transmission is as much an imperative as any government AI regulation.
Regulation: Moving from Voluntary to Mandatory
Canberra has chosen a staged, risk-based path. A voluntary safety standard issued in September 2024 sets guardrails around accountability, human oversight and transparency. However, voluntary standards and self-regulation in banking, mining, aged-care, construction and advertising have, at best, a mixed record in Australia.
In December 2024, the government explored tougher mandatory rules for high-risk use cases, and ministers have agreed on the need for national ethics guidelines for public sector deployments. The Australian Framework for Generative AI in Schools is now guiding states and territories regarding safety, transparency, fairness, privacy, accountability and human oversight.
The government is moving, but too slowly and without a comprehensive strategy. Policy is piecemeal and support ad hoc. Federal funding now spans more than a dozen grants and pilot schemes, from AI Adopt Centres to medical research funds, without a single point of accountability.
A clear and coherent national plan for policy development is desperately needed. One that is not naïve about the risks, but balances this with incentives for business and other institutions.
Research and Funding: Much More is Needed
The 2024-25 Budget earmarks $ 39.9 million for policy capability and a national advisory body, whilst $ 17 million backs AI Adopt Centres that help SMEs. Health research gained $ 29.9 million for AI diagnostics in 2024, and regional grants of up to $ 0.5 million encourage practical deployments.
Other measures are being taken. December 2024 saw the opening of the Responsible AI Research Centre in Adelaide to tackle explainability and hallucination risks. The aim is to position Australia at the frontier of knowledge, understanding and safety.
However, the dollar amounts are inadequate, and the allocation is piecemeal. But funding for research and development should not only be left to governments. Business and unions must be encouraged to co-invest with universities and research centres such as the CSIRO. This is in everyone’s interest.
Conclusion: The Need for an Artificial Intelligence Summit
The Hawke-Keating economic reforms were not without their mistakes, and sectors such as manufacturing were badly hurt. However, the long-term benefits are unarguable: floating the dollar, deregulating finance, cutting tariffs, rewriting wage setting, and unleashing competition made Australia more productive, more resilient and richer.
The transition to an AI-driven economy will not be without its own pain points. It is nonetheless inevitable, so the imperative is to give Australia the best chance of success. This requires a national consensus.
Business needs clarity on policy, supportive taxation, financial incentives and skilled workers. It also needs to be encouraged to do its part in training and investment.
The energy grid requires upgrades to provide an increased supply of reliable energy.
Clear worker protections will be required to keep labour engaged and supportive. Unions also need to play their role in investment in training and supporting skilled migration.
Education and training, from primary school through to professional development, must evolve rapidly to keep pace with AI technology and its applications. Educators and institutions need rapid reskilling and to be given incentives to make the necessary changes.
Research requires more funding from public and private sources.
State and Federal governments must support sectors that require it.
The Federal Government needs to accelerate policy and regulatory development.
Only Canberra can provide the means to achieve a national consensus where all sectors cooperate and coordinate.
OCNUS Consulting recommends a National AI Summit as soon as possible. Accords of various kinds need to be forged. Without these formal compacts, maintaining national consensus will be difficult. If we miss this opportunity to prepare the economy for the AI revolution, then, at a minimum, Australia risks repeating the post-mining-boom drift, where opportunities for national reinvigoration were squandered. At worst, Australia could quickly fall behind and face a fate like twentieth-century Argentina.
The Hawke-Keating lesson is clear: bring all parties into one room, agree on a national AI strategy, share the risks and spread the rewards.