I acknowledge the Ngunnawal people, traditional custodians of the land on which we meet, and acknowledge all First Nations people present.
My thanks to Chifley’s multitalented Executive Director, Emma Dawson, for bringing us together today at the end of Budget week for this panel on ‘Innovation, Productivity and Prosperity for All’. It is a pleasure to open the session before we hear from the 4 brilliant panellists: Kate Cornick, Tony Sheldon, Amit Singh and Rob Nicholls.
A paddock revolution
Let me start with a story.
For thousands of years, farming began by turning the soil over.
The plough was one of humanity’s oldest technologies. It broke the ground, buried weeds and prepared a seedbed. It was so familiar that most people would scarcely have thought of it as a technology at all. It was just farming.
Then, in the second half of the twentieth century, some Australian grain growers began asking a provocative question: what if we stopped?
What if the best way to grow a crop in a dry country was to disturb the soil less? What if farmers could sow directly into the ground, keep stubble on the surface, reduce erosion and conserve moisture? What if progress meant doing less to the soil, rather than more?
At first, zero‑till looked like a very antipodean kind of innovation: less big‑screen announcement, more paddock trial. Less breathless futurism, more farmers standing beside seeders asking whether the thing would work after 3 dry months and a nor‑westerly.
Farmers experimented. Scientists worked with them. Machinery had to change. Practices had to be adapted to different soils, rainfall patterns and weed pressures. Growers learned from growers. Field days, trial plots and farmer groups did the slow work of turning an idea into a system.
The Australian story of conservation agriculture was an interplay of policy, economics, science and farming (Bellotti and Rochecouste 2014). Australian rain‑fed agriculture faces old, infertile soils and low, variable rainfall. Farmers also operate in export markets with slender subsidies. Those conditions created a persistent search for ways to lift yields per millimetre of rainfall and dollars per hectare.
Conservation agriculture offered a practical answer: minimum or zero tillage, retention of crop residue to protect soil cover, and rotations across cereals, oilseeds and pulses. Farmer experimentation began in the 1960s. By the late 2000s, zero‑till adoption in winter cereal regions was 80 to 90 per cent. One account described it as a 40‑year overnight success (Bellotti and Rochecouste 2014).
The innovation has diffused beyond Australia. The Australian Centre for International Agricultural Research is working with partners in India, Nepal and Bangladesh to help smallholder farmers adapt conservation agriculture to their own conditions: smaller farms, different machinery, different crops, different water pressures, different lives (ACIAR 2019).
An idea changes the world when people can test it, adapt it, trust it and use it.
The dividend comes when ideas travel
That is the theme of this speech: the diffusion dividend.
Australia should care deeply about invention. We should back discovery, research and enterprise. We should celebrate the people who push out the frontier of what is possible. By better targeting the Research and Development Tax Incentive and expanding venture capital tax incentives, the budget strengthens the incentive to create new knowledge.
But invention forms only part of our focus. A substantial economic prize comes when good ideas travel. Productivity rises when better tools and smarter ways of working spread from leading firms into everyday workplaces. Innovation becomes powerful when it stops being exceptional and starts becoming ordinary.
That is true in agriculture and manufacturing. It is true in health, education and the care economy. And it is true of artificial intelligence.
The central question is what Australia can invent, and how widely we can spread the gains.
Workers benefit from innovation when it reaches their workplace, strengthens their skills, improves their jobs and lifts their wages. A breakthrough in a lab is valuable. A breakthrough that helps a nurse, a teacher, a technician, a farmer or a small business owner do better work is more valuable still.
That is the diffusion dividend.
It is a productivity idea. It is also a Labor idea.
Productivity is about possibilities
As the Assistant Minister for Productivity, I’m the first to admit that ‘productivity’ is sometimes treated as a cold word. Output divided by hours. The sort of term that makes even the best journalist wonder whether there is a synonym that won’t make readers flip the page.
But productivity is really about possibilities.
It is how living standards rise. It is how we afford better aged care and stronger schools. It is how families gain higher incomes and more time.
A society with rising productivity has more choices. A society with stagnant productivity finds every argument harder. Wages are harder to lift. Public services are harder to fund. The politics of scarcity creeps into places where generosity should live.
So the productivity question is one of the central questions for Labor in government. How do we build an economy that is dynamic and fair? How do we make sure growth expands what Australians can do with their lives?
Diffusion is a critical part of the answer.
Innovation is often portrayed as a lightning strike: a genius in a garage, a scientist in a lab, a founder with a pitch deck and a suspiciously confident hockey‑stick graph.
Those stories can be true. Societies need invention. We need discovery. We need ambition. We need research.
Yet the economic history of innovation is also a history of spread.
Electricity did much more than add a new source of power. It changed the layout of factories and the timing of work. Computers did much more than land on desks. They changed design, logistics, jobs and communication. The internet did much more than connect machines. It reshaped commerce and public administration.
The invention is the opening chapter. Diffusion is the long middle of the book.
And that middle is where many of the gains are won or lost.
When technology travels unevenly
Recent firm‑level data from France shows the importance of adoption. It finds that companies adopting AI are hiring more (Aghion et al. 2025). AI adoption was associated with increased employment and sales, including in roles previously thought vulnerable to automation.
In fact, the biggest employment risk from AI may be working for a business that fails to adopt it and then falls behind. Think of it as the equivalent of working for a boss who insists that the overhead projector will make a comeback.
The French research also finds something that is likely to be true in Australia too: larger firms are more likely to integrate AI, which risks widening existing divides.
This is a worker issue as much as a technology issue.
A worker benefits from a new technology when their workplace has the capacity to adopt it well. That means training and trust. It means redesigning jobs with workers, rather than dropping software onto them from a great height and calling the process transformation.
It also means recognising that workers are sources of innovation themselves.
Some of the best ideas in any workplace are held by the people closest to the task: the nurse who knows where the delays occur, the mechanic who understands which faults keep returning, the teacher who knows which digital tools improve learning and which merely distract.
Too many organisations underuse that knowledge. They treat innovation as something imported from outside, rather than something drawn from within. They buy the system, hold the webinar, send the login details, and wonder why productivity has failed to rise by the end of the month.
Diffusion done well is more human than that.
It involves translation and adaptation. It involves building confidence and capability.
The extension instinct
Let’s come back to agriculture for a moment.
Agriculture has long understood that research and adoption are different tasks. Discovering a new crop variety is one achievement. Helping farmers decide whether it suits their soil, climate, machinery and finances is a separate achievement.
Institutional research is one step. Farmers then evaluate an idea on their own land. Early adopters share experience at field days and through local networks. More farmers implement the practice, adapting it to local conditions. Farmers may even identify fresh problems and suggest them as research priorities.
That is diffusion as a living system.
Zero‑till spread because farmers saw trial plots on land that looked like their own. They watched seeders handle stubble. They compared moisture retention, erosion control and yields. They learned from neighbours who had tried it first. Departmental officers worked with farmers on demonstrations. Local manufacturers adjusted planters. Agronomists and farmer‑advisers translated general findings into local practice.
Successful farmer‑led diffusion is site‑specific, because farms differ in soils, weeds and climate. It is integrated, because farmers test technologies inside whole production systems. It is incremental, because large changes often arrive through smaller adjustments. It is persuasive, because farmers communicate risk and uncertainty in the language of people whose livelihoods are on the line.
That is the extension mindset.
As we think about artificial intelligence, robotics, advanced manufacturing and digital services, we should ask: where are the extension systems of the modern economy? Who helps a small business make sense of a new tool? Who helps workers gain confidence with new systems? Who helps firms learn from each other? Who turns a promising technology into a practical improvement in a real workplace?
Those questions are less glamorous than talking about the next billion‑dollar AI startup. They may be more important for living standards.
Because the average worker is unlikely to be employed by a firm that appears on the cover of a technology magazine. The average worker is in a school, hospital, warehouse, workshop, public agency, aged care centre, supermarket, construction site, office or small business.
If innovation policy focuses only on the most advanced firms, it will miss most of the economy. At present, just 7 per cent of small and medium enterprises say that they are making broad use of AI (National AI Centre 2026). Thanks to our decision this week to make the instant asset write‑off permanent, these firms now have a stronger incentive to invest in new technology.
The real challenge is to keep backing the cutting edge, while bringing the rest of the economy closer to it.
A four‑part diffusion agenda
Let me suggest 4 ways to ensure technological diffusion forms part of a progressive productivity agenda.
1. Put workers in the cab
Technology should strengthen human judgment and extend expertise. In health care, AI can help nurses and doctors manage information and read scans. In logistics, it can improve routes and reduce fatigue. In public services, it can simplify paperwork and free frontline staff to spend more time with the people they serve.
The best use of technology is as a complement to workers: a lever for expertise, rather than a replacement for it.
That requires voice. Workers should have a say in how new tools are introduced. The people who know the work should help shape the tools that change the work.
This is sound economics. Workers know which tasks waste time. Workers know which systems duplicate effort. Workers know where errors happen. Workers know which innovations will solve a real problem and which will only give the old problem a fresh dashboard.
A diffusion agenda that trusts workers will be stronger and smarter.
2. Build skills and open gates
Diffusion depends on people who can adapt. We often talk about training as though it were a repair shop for those left behind by change. That is the wrong frame. Skills are economic infrastructure. They are what allow people to use new tools, move into better jobs and bring fresh ideas into the workplace.
If we want technology to raise wages, we need workers who can use it confidently. If we want artificial intelligence to complement expertise, we need to invest in expertise. If we want the gains from innovation to spread, we need skills that spread too.
The old model imagined education early in life, work for several decades, then retirement. That model now looks about as current as a dial‑up modem. People will need to learn across their working lives. Sometimes that will mean formal qualifications. Sometimes it will mean short courses, workplace training, peer learning or industry‑specific support.
Mobility belongs here too. When workers can move more easily between jobs, ideas move too. Skills spread. Better employers attract talent. Poorer employers have to improve. Labour mobility helps match people to roles where their abilities are used well.
That is one reason our government is scrapping non‑compete clauses for everyone earning less than $183,000. A labour market where people are trapped is a labour market where knowledge moves slowly. Job mobility boosts dynamism and innovation.
3. Honour the translators
The extension lesson from agriculture is clear: useful ideas need channels. They need trusted messengers and practical demonstrations. They need peer networks and feedback loops.
In the broader economy, that means taking seriously the institutions that sit between invention and adoption. Industry advisers. Trainers. TAFE teachers. Workplace delegates. Managers who know how to organise change. Public servants who can make systems work. Sector bodies that can share lessons with something livelier than an eye‑glazing strategy document.
The translators rarely get the glory. They are essential.
This is particularly important for small and medium‑sized businesses. A large firm can hire a team to experiment with AI, new software or advanced equipment. A small firm may have the same opportunity in theory, and a much harder path in practice. The owner is doing payroll on Tuesday, chasing invoices on Wednesday and discovering on Thursday that the printer has joined a separatist movement.
Diffusion policy is about meeting firms where they are.
4. Make public institutions diffusion engines
Government is more than a regulator standing outside the economy with a clipboard. government buys services, builds infrastructure, funds research, educates workers, designs markets and delivers public services.
When government adopts technology well, it can improve services and build confidence. When government adopts technology poorly, it can produce a magnificent new portal that requires people to print a form, scan it, upload it, and then ring someone to check whether it arrived.
We can do better than that.
Public institutions should help proven ideas travel. They can gather evidence, test what works and spread practical lessons. They can use procurement to encourage better systems (Charlton 2026). They can support common standards, open data, interoperability and digital foundations that lower the cost of adoption. And in the case of special investment vehicles like the National Reconstruction Fund, they can make investments that crowd in private capital and encourage the diffusion of ethical AI.
This is also where fairness must be built in. The benefits of innovation should be shared. That happens through choices, institutions and rules. New technologies can raise productivity while leaving wages stagnant if workers lack bargaining power. They can make work safer or more stressful. They can create flexibility for workers, or flexibility only for employers.
The goal should be clear: technology should raise standards, widen opportunity, strengthen dignity at work and lift wages.
Dignity can sound abstract, but anyone who has had a bad job knows exactly what it means. It means having some control over your day. It means being treated as a person rather than a cost centre. It means your skills being recognised. It means technology helping you do better work, rather than making your work more frantic, more monitored, more precarious or more disposable.
Diffusion and fairness belong together.
When good ideas spread through strong institutions, skilled workplaces, fair rules and worker voice, productivity gains can become wage gains. When diffusion occurs in an economy with weak bargaining and concentrated power, the gains can pool at the top.
Labor’s task is to shape diffusion so it serves working people.
Productivity with purpose
This connects directly to the Chifley tradition.
Ben Chifley’s economic vision was always about growth with a social purpose. It was about building a country in which the economy served human ends: security, opportunity, dignity and the chance to build a decent life.
Today, that tradition requires us to think seriously about productivity.
Progressives sometimes talk about productivity defensively, as though the word has been permanently captured by people who think that Australia’s route to success lies in taking the low road: cutting worker protections, cutting wages, cutting environmental protections and cutting the social safety net.
A progressive productivity agenda is about investing – in individuals, in institutions and in infrastructure. We choose the high road to productivity growth, and we recognise that productivity is fundamental to the Labor project.
Productivity is how we make room for ambition. It is how we fund the social wage. It is how we improve wages while easing inflationary pressure. It is how we build a country capable of meeting great challenges, from the care economy to digital transformation.
The progressive productivity agenda is a long way from austerity or managerial slogans. It is about helping people work better, with better tools, better skills, better systems and better institutions.
The dividend we choose
Zero‑till farming changed Australian agriculture because it travelled. Farmers tested it, adapted it, shared it and improved it. Scientists helped. Machinery changed. Local knowledge did the rest. The idea became valuable because it diffused.
That is the lesson for our time.
Australia should keep backing discovery. We should keep celebrating the researchers, entrepreneurs and workers who create new possibilities. But the next productivity dividend will depend on how widely those possibilities spread.
The test of an innovation system is whether it produces brilliant ideas and whether those ideas reach the people who can use them.
Whether they strengthen work.
Whether they lift wages.
Whether they expand opportunity.
Whether they help build a fairer, more prosperous Australia.
That is the task before us.
That is productivity with purpose.
That is the diffusion dividend.
References
Aghion, P., Bunel, S., Jaravel, X., Mikaelsen, T., Roulet, A. and Søgaard, J., 2025. ‘How different uses of AI shape labour demand: evidence from France’, AEA Papers and Proceedings, 115, pp. 62–67.
Australian Centre for International Agricultural Research, 2019. Media note on zero‑till and conservation agriculture, ACIAR, Canberra.
Bellotti, B. and Rochecouste, J. F., 2014. ‘The development of Conservation Agriculture in Australia: farmers as innovators’, International Soil and Water Conservation Research, 2(1), pp. 21–34.
Charlton, A., 2026. ‘Sliding Doors: the critical window to back Australian AI’, Speech at the University of Technology Sydney, 1 May.
National AI Centre, 2026. ‘AI Adoption’, available at https://www.ai.gov.au/news‑and‑insights/ai‑adoption‑tracker/ai‑adoption