This is part four of a four-part article series. Read parts one, two and three.
Universal Basic Intelligence (UBIQ) begins with a simple proposition: every Australian should have access to a meaningful foundation of capability and agency in a society increasingly shaped by machine intelligence.
But declaring intelligence a universal public good does not tell us how that capability will grow.
Governments may be tempted to reach for familiar instruments: a national AI portal, subsidised access, public information campaigns, short courses, training entitlements or centrally delivered services. These measures may extend access, and some will have an important role to play. Yet access alone does not produce agency. A society can distribute technology widely while leaving the capacity to shape, direct, and benefit from that technology concentrated in the hands of a few.
The challenge before Australia is therefore larger than digital inclusion. The question is not simply whether every Australian can access and use AI. It is whether every Australian and every community can develop the capability to think, create, deliberate, and act with machine intelligence in ways that strengthen their own futures.
Universal Basic Intelligence is the public commitment. Community capability is how we realise it. Co-intelligence is the outcome.
The society we are trying to build
The goal is not universal AI adoption. Nor is success simply an increase in training completions, prompts written, software licences issued or public services automated.
The deeper ambition is a society in which people and communities can think boldly and independently about their own purposes, use machine capability without surrendering human judgement, and participate in shaping the intelligence systems that increasingly influence their lives.
A co-intelligent society should multiply ideas, knowledge and culture rather than narrowing them. It should strengthen democratic participation, enrich daily life and wellbeing, expand imagination, and distribute power more widely. Most importantly, it should enable communities to define and pursue purposes of their own rather than becoming passive recipients of technologies and priorities designed elsewhere.
This makes UBIQ ultimately a question of civic capability and power. Who decides what intelligence is used for? Whose knowledge matters? Who retains the ability to exercise judgement? Who gets to create, rather than merely consume?
The architecture we build will shape the answers.
The final mile is not access
Much of Australia’s AI discussion currently focuses on access, adoption, and literacy. How do we ensure people can use the tools? How do we increase uptake? How do we teach people what AI can do?
These are important questions, but they are not sufficient.
History repeatedly shows that access and capability are not the same thing. People can receive services without gaining power. They can consume technology without understanding or shaping it. They can participate in programmes without strengthening their ability to act independently once the programme disappears.
Imagine two Australians with access to exactly the same AI model.
One works in a professional environment where experimentation with digital tools happens during paid work. Their employer provides enterprise software, technical support, training, and colleagues who continually exchange new practices. Over time, that person may develop personalised AI assistants connected to their information, workflows, and professional knowledge.
The other encounters AI through a free consumer account, without institutional support, spare time, trusted guidance or a community of people experimenting alongside them.
Technically, both have access. Practically, they inhabit entirely different intelligence environments. The risk is that Australia solves the access problem while leaving this agency gap largely untouched.
That is why the final mile is not access. It is agency: the confidence, judgement, relationships, knowledge, and practical capability required to use machine intelligence in pursuit of purposes people have chosen themselves.
Universal does not mean individualised
One of the central lessons of public policy is that collective challenges can be weakened when they are translated entirely into individual solutions.
Individual entitlements can provide essential support. They protect rights, improve access, and deliver necessary services. But capability develops differently. It emerges through relationships, practice, experimentation, participation, and shared learning.
When relational challenges become individual transactions, public effort can become dominated by administration and service delivery while community capability remains unchanged.
The lesson for UBIQ is important. Universal Basic Intelligence should not simply become an AI voucher scheme, a subsidised software subscription or an individual training entitlement. These interventions may improve access while leaving the underlying agency gap untouched.
The principle should be that UBIQ is universal, but capability is built collectively.
Co-intelligence cannot simply be delivered as a product. It develops through shared practice.
Australia’s community capability gap
Earlier in this series we described an emerging agency divide between people able to direct and shape intelligence systems and those increasingly subject to them. There is a corresponding policy challenge: Australia’s community capability gap.
Today’s AI agenda understandably concentrates on frontier research, national compute, data centres, regulation, corporate adoption, productivity, and AI literacy. All matter. But they can leave a more fundamental question unanswered: can communities use and shape machine intelligence for purposes they determine themselves?
White-collar workers have legitimate concerns about how AI will reshape jobs, professions and identities. Yet they are also comparatively likely to encounter AI through paid work, enterprise platforms, professional networks, and organisational training.
An AI strategy designed primarily around their experience risks overlooking people whose work, knowledge, and contribution sit outside corporate settings.
When productivity becomes the dominant measure of AI value, corporate and administrative purposes become highly visible. Community, cultural, civic, and relational purposes become harder to see: preserving a town’s history, helping carers navigate fragmented systems, strengthening a language, improving local deliberation, understanding why young people are disengaging, supporting environmental stewardship or simply helping a community imagine a different future.
These are also legitimate uses of intelligence.
The policy challenge is therefore not only to improve adoption. It is to increase communities’ capacity to act.
Government must choose its role deliberately
Building that capability requires a broader conception of government’s role than simply providing an AI service.
Government may need to act simultaneously as sense-maker, champion, convener, regulator, provider, facilitator, funder, and learner. These roles are not mutually exclusive, and direct provision will sometimes be essential.
But if UBIQ is intended to distribute agency rather than dependency, government’s defining posture should be enabling.
Government can help Australians understand possible futures through foresight, public dialogue, and accessible information. It can articulate a positive public vision for AI that extends beyond productivity and risk. It can fund experimentation, establish safeguards and open protocols, make public information more usable for community purposes, and support the infrastructure required for communities to participate meaningfully.
It should also learn from what communities discover rather than assuming that the centre already knows the destination.
Government cannot give people agency. Agency cannot be administered into existence. What government can do is create the conditions, resources, protections, and infrastructure through which people build agency together.
Start where trust already lives
The encouraging part is that Australia does not need to invent the social infrastructure required to do this. Much of it already exists.
Across the country are neighbourhood houses, public libraries, First Nations community-controlled organisations, multicultural and faith organisations, arts institutions, unions, professional associations, local councils, regional organisations, community health providers, and countless informal networks.
These institutions may not look like AI infrastructure. But they possess something no national AI portal can manufacture: relationships.
They understand the local context. They possess social legitimacy. They know how people actually learn from one another. They can create spaces where uncertainty is acceptable, experimentation is social, and people can work on problems that matter to them rather than problems selected by a technology provider.
Australia’s most important future AI infrastructure may therefore not be another data centre. It may be the relational infrastructure through which people develop the confidence, judgement, and collective capacity to use intelligence for purposes of their own.
A network of Community Co-Intelligence Hubs
One practical way to begin would be to establish a nationally supported network of Community Co-Intelligence Hubs, embedded within institutions that communities already know and trust.
A hub would not necessarily be a new building or organisation. It could be a capability hosted within a neighbourhood house, public library, First Nations organisation, arts centre, regional organisation or community service.
Nor would these be technology centres where experts demonstrate the latest AI tools. They should not become conventional training programmes organised around a particular platform.
They would be welcoming, ongoing places and networks where communities explore how human knowledge and machine intelligence can work together to pursue purposes they have chosen themselves.
A regional town might explore alternative economic futures. Carers could find new ways to reduce the administrative burden surrounding care. A cultural organisation could preserve and govern local memory. Residents could make sense of hundreds of community experiences. First Nations organisations could experiment with machine capabilities while maintaining authority over their knowledge. Communities could strengthen environmental stewardship, improve local deliberation or imagine possibilities that neither government nor industry had thought to commission.
The objective would not simply be to produce successful AI applications.
It would be to leave communities more capable of framing questions, governing knowledge, exercising judgement, challenging machine outputs, imagining alternatives, and acting collectively.
That distinction changes how these hubs should be designed. They should begin with community purpose rather than technology; build on existing relationships rather than bypassing them; recognise lived, vocational, and cultural knowledge as genuine intelligence; allow knowledge holders to determine what can be shared; keep human judgement consequential; and support learning over time rather than cycling through short-term pilots.
Local authority, national learning
A distributed model could operate at three interconnected levels. Locally, trusted organisations would host co-intelligence activity around locally defined purposes.
Governments and specialist partners would provide enabling infrastructure: funding, safeguards, accessible public information, technical support, interoperability standards, and appropriate access to technology.
Nationally, these local efforts could form a learning network through which approaches, governance patterns, failures, and lessons travel between communities.
But the knowledge itself would not automatically need to travel. This is crucial.
A community in Arnhem Land and a neighbourhood organisation in Western Sydney may develop very different approaches to AI. The purpose of national architecture should not be to standardise them into a single model.
It should allow learning to travel while authority remains local. Methods can travel. Protocols can travel. Lessons can travel. Technical patterns can travel.
Local knowledge does not need to be extracted and centralised simply because useful learning has occurred. That is what a genuinely distributed intelligence infrastructure could look like.
Measure agency, not usage
This approach also demands a different definition of success.
Technology programmes typically gravitate towards measurable adoption indicators: how many people participated, how many licences were issued, how many tools were used, how much time was saved, and what economic benefit was generated.
These measures can be useful. They tell us surprisingly little about whether people have gained agency. A co-intelligence agenda would also examine changes in confidence, participation, local decision-making capability, knowledge sovereignty, collective learning, imagination, dissent, and community power.
The most important evaluation question might be much simpler:
Is this community now more capable of shaping its own future?
That is a more demanding test than adoption. It is also much closer to what Universal Basic Intelligence is intended to achieve.
Australia as a living demonstration
Australia has an opportunity to become more than an adopter of emerging technologies. It can become a demonstration of how a diverse democracy builds intelligence infrastructure around human flourishing, participation, and community capability.
Doing so will require investment not only in technical systems, but in relational ones. Not only in compute models, but in trust, participation, and collective capability. Not only in helping Australians use artificial intelligence, but in ensuring they can contribute to, challenge, and shape the intelligence systems forming around them.
The choice is not whether AI becomes part of Australian life. That transition is already underway. The choice is whether machine intelligence primarily amplifies the purposes and power structures that already dominate our institutions, or whether it helps Australians create new forms of agency, culture, participation, and collective possibility.
Universal Basic Intelligence is the public commitment. Community capability is how we build it.
Co-intelligence is what becomes possible when people combine human judgement, local knowledge, and machine intelligence in pursuit of purposes they have chosen for themselves.
UBIQ establishes the floor. Community capability allows Australians to build above it.
The real measure of success will not be how many Australians use AI. It will be how many gain greater power to shape their own futures.
