A friend of mine who is very interested in the effects of AI on jobs, especially for young people, sent me a message pointing to a new post on LinkedIn. The post was to a new Australian Government initiative:

I think the separation between human and AI skills is meaningless as a foundation for understanding work now. My starting proposition is that most human work can be assisted and empowered by AI models, and a skills system needs to recognise the capability that creates.

My immediate reaction was to ask both ChatGPT and Claude to review the ‘insight’. Both the AIs identified the same problem I had, but ChatGPT said it best:

The definition may be too human-exclusive for an AI-enabled workforce. In many jobs, capability is already becoming human + AI: software developers working with coding agents, radiologists using AI-assisted image analysis, and analysts using AI to research, model and draft at much greater speed. A future skills system needs to measure what people can achieve with AI, not just what they can do unaided.

Are the people writing these policy documents using the latest versions of AI models?

I really wonder if people who think that human skills can be considered separately from human AI assisted skills are using the latest models - like ChatGPT 6.0 or even Claude Opus 5.0. I know that you need paid subscriptions to access the most recent models from OpenAI and Anthropic which is hardly a barrier for a national policy agency. So I have to ask: have the people developing this policy used the latest models? If not, they have no real sense of what these models can do, and human-exclusive skill looks like a reasonable starting point. It isn't.

I feel strongly that this type of thinking from people developing government policy is both wrong and, just as importantly will become more and more out of date as the capabilities of AI models keep improving. What we should be focusing on is what are the new skills and capabilities that AI models provide us and any list of these skills and capabilities needs to be updated after each major new model release.

This is another example of a fork in the road. The old paradigm of careful policy development outlined by Jobs and Skills Australia is just not relevant to the new future we have entered where AI models can allow anyone who is prepared to invest some time in using them to do things that they were never capable of doing before.

A common skills language will only work if it is trusted by the people and institutions that use it. That requires careful governance, quality assurance, practical testing and genuine engagement with the sectors that will adopt it.
Megan Lilly is Acting Commissioner, Jobs and Skills Australia.

This sounds good but is divorced from the reality where my skills and capabilities increase at least 3X every time a major new AI model is released. Where does that fit in a system built around defining skills as exclusively human?

How I built a new immigration simulation in one day

I am very interested in the major risk factors that affect resilience and demographic declines in developed economies affects two of the key resilience risks most developed economies are facing:

• Rapidly ageing populations

• Deteriorating social cohesion due to high levels of immigration

I have been using my AI models to research how rapidly populations are ageing in OECD and G20 countries to get a better understanding of the scale of this problem - more on this soon in a follow on post. Using the research that had been completed, I got Codex (reviewed by Claude) to develop an Australian Demographic Simulation

This demographic simulator was developed by Codex this morning. I asked Codex to make some minor changes to the simulator and then got Claude to review it for accuracy.

I will get Codex to use my automated publishing system and push it to my GitHub site where I will review it, before making it live. This whole process from signing off on the final version of the simulation to going live takes me ten minutes of my time. Is this a skill that will be tracked going forward?

Conclusion

I think this approach to defining skills is completely flawed. It makes human exclusivity the organising principle when the capability we need to understand is human and AI together. The test I would put to Jobs and Skills Australia is straightforward: show how this system recognises someone whose thirty years of experience can now produce outcomes that were previously beyond their reach.

If you are already doing work with AI that you could not do before, what would a skills assessment need to capture to recognise that change?

Subscribe to Our Resilient World

I have crossed a threshold I had been trying to reach for eleven years. Using the AI Product Flow Guide and the Product Owner workflow I first tested through Kangentic, I can now create a working prototype as a sole Product Owner even though I cannot program. The AI does the implementation work; I define the purpose, boundaries and acceptance decisions. (more about the AI Product Flow guide soon).

That does not mean I can yet build a complicated production product on my own. My current judgement is that it will probably take two more major updates to both Codex and Claude before a non-programmer Product Owner can lead that work without a programmer in the delivery path. Even then, security, legal, operational and specialist domain expertise will still be required where failure could harm people.

This two major updates is my best guess. It is based on the distance the models and tools travelled between my failed attempts in 2023 and 2024 and the working products I accepted in 2026. To explain why I draw the line at a working prototype, I need to start at the beginning.

In 2015 I could define the product but not build it

In February 2015 I wrote a business proposal for a proof of concept called Foreign Fighter Detect for Daon, a biometrics company I had worked for. The idea was to use more than facial recognition: height, stride length, clothing, weapons, gestures and other information in video that would contribute to a composite identity, with a human expert making the judgement.

The proposal relied on the computer-vision technology available at the time and needed developers plus specialist object-recognition and neural-network expertise. Daon judged it too technically risky and outside its strategy. I could see the opportunity and define a bounded proof of concept, but I could not build it without a technical team.

ChatGPT could write the plan, not deliver the product

I began using ChatGPT 3.5 in December 2022 and soon moved to GPT-4. I had spent much of my career translating between a business need and an IT team, so the promise was obvious: perhaps I could describe the outcome and let the AI write the code.

In 2023 I used GPT-4 to define a university AI platform, an academic-feedback application and a climate and resilience “digital brain”. It produced features, user stories, risks, a steel thread and an MVP in minutes. Those were useful product artefacts, but they were not working products.

The missing work returned to me. I had to decide where files belonged, install dependencies, interpret errors, judge whether tests proved anything and work out whether a repair had broken something else. By April I wrote that I felt as though I was on a roller-coaster and wanted to get off.

Auto-GPT did not solve the problem. It repeated failed approaches, consumed API calls and once wanted to install software on my computer, so I stopped the run. I learned that an agent able to make another attempt does not necessarily understand why the last one failed.

xAlerts exposed the work a programmer had been doing

In May 2024 I tried again with xAlerts - an updated version of my idea I pitched to Daon back in 2015. The first steel thread for the demo I was trying to build was simple: select a video, extract frames, send them to GPT-4o, look for a handgun or rifle and generate an alert for a person to review. My instruction was explicit: “GPT4o would write all the code using a test driven development approach.”

There was progress. I installed the tools, displayed “Hello, xAlerts!”, extracted frames and completed the frame-capture feature. There were unit tests, mocked model calls, a web form and attempts at the full workflow. There were also many hours of failure. Modules could not be found, tests failed during collection, replacement code reproduced old errors, and integration changes damaged functions that had already passed their tests. I became the courier carrying errors from VS Code back to ChatGPT—what I came to call copy-and-paste hell.

A programmer friend spent about an hour repairing the demo and got it working. I had proved the concept, but I had not proved that a non-programmer could deliver it. Removing the human programmer from the conversation had not removed the release management, source control, integration and debugging work the programmer had been doing. It also demonstrated that the xAlerts demo was now so complicated that I could not change any of the functions without breaking something else.

Better models still needed a better system

By early 2026 the models were much better. A Living Well Health Tracker reached ten iterations, 143 tests, a dashboard, local saving, import and export, reference ranges and trend views. Then a link audit found eight PubMed identifiers pointing to the wrong papers, one wrong PMC identifier and other citation problems.

The code could work while the evidence shown to the user was wrong. Tests proved that the code did what it had been told; they did not prove that what it had been told was true. That distinction matters in any product, and it becomes decisive in health, identity or public services.

My Knowledge Management work exposed a related failure. Individual components worked, but integration broke them and I still had to move context between chats, manage sessions and correct the operating process. Faster models had accelerated the work without giving it a reliable source of truth or a dependable path from one stage to the next.

The change was the workflow around the models

I stopped treating the chat as the product workspace. The product moved into a Git repository where an agent could read the governing files, make a bounded change, run tests, preserve evidence and return the judgement decisions to me.

The AI Product Flow Guide supplied the missing structure. I start with the product intent, an explicit boundary, a named source of truth, a steel thread (the steel thread is the thinnest useful real path through the product or workflow) before deciding if I want to proceed and create the MVP. With every step the artefacts that are created before the agent begins coding becoming the AI harnesses that the agent works within.

Kangentic helped me test the multi-stage Product Owner workflow, but the tool did not survive unchanged as the execution engine. It now tracks Backlog, To Do and Done, while Codex performs the work in an isolated repository worktree and Claude is used to test the first version of the steel thread or MVP. Only when Claude finds no significant issues do I review the product (can be code, documents, research etc - all follow the same basic workflow). Getting to this stage where my role is reduced to that of a Product Owner - defining purpose, reviewing and approving artefacts like customer journey maps, roadmaps and a workflow simulation and sample graphic user interface was difficult.

It took me many sessions with both Codex and Claude to get them to do the work and not require me to view every step in the process. It seemed to me that I was breaking the pattern that both these AI models had been taught to follow. The result is a sea change for me - I am not copying and pasting between chat sessions and codex or claude. I get links to the folders I am working on in VS Code instead of drilling down many levels to find the latest document or product I need to review. My problem now is that it is so easy to use my Kangentic workflow and I still have to wait for the agents to work through the coding processes I run too many sessions concurrently. I am the ‘flow control’ layer and I have to do less at the same time to achieve more.

The working-prototype threshold has been crossed

The Portfolio Dashboard is the clearest software example. I defined a private, read-only starting page for my current work, repositories and tools. Codex built and tested the bounded MVP; I opened it through the normal desktop path, judged whether it was useful and accepted it.

The software still had to be programmed, tested, debugged and installed, but another human programmer was not in that implementation path and I did not have to pretend to be one. I could remain the Product Owner. The dashboard has since been improved through the same bounded method.

The broader 2026 record gives me more confidence that this was not one lucky result. The same controls have supported completed research products, simulations and publication packages. Those outputs do not prove market demand or production fitness, but they do show repeated movement from intent to a completed product within an agreed boundary.

Production is a different threshold

A working prototype proves that the product idea can be made real and tested by its Product Owner. A production product must also survive sustained use, hostile inputs, security threats, upgrades, monitoring, support, legal obligations and failures that may affect other people. My 2026 evidence does not cross that threshold.

The remaining problem is not simply that Codex and Claude sometimes write faulty code. They still lose context, repair the wrong layer, treat technical permission as a Product Owner decision and accept tests that prove the wrong thing. The workflow contains those failures, but containing them still takes too much operating design and specialist checking for a complicated product.

My forecast is that two more major Codex and Claude updates could close enough of this gap for a non-programmer Product Owner to lead a production build. I expect better integration reasoning, more reliable long-running state, stronger independent checking and fewer false handoffs. I do not expect an AI model to remove the need for human responsibility or specialist expertise.

What has changed

In 2015 I needed a development team before I could test the product idea. In 2024 I could make parts of a product, but I still needed a programmer to recover the build. In 2026 I can use the AI Product Flow Guide and the workflow developed through Kangentic to take a bounded idea through implementation, testing and Product Owner acceptance as a working prototype.

That changes who can start. A person who understands a problem can now build enough of the solution to test it, improve it and show a future technical team what “good” looks like. The prototype also leaves behind the product intent, decisions, tests and evidence that a production team can build from.

The next question is no longer whether AI can write code for a non-programmer. It is whether the next model generations can make the path from a Product Owner’s accepted prototype to a production product reliable enough to trust.

I am really excited about the power that AI models have given me. I had so many ideas that I wanted to implement but without the technical skills to do it myself I was always dependent on someone else. Now I can do a lot of what I want and I am confident that in a couple more generations of models if the progress continues I will be able to build the One World System that I wrote about in 2022. I will keep updating this journey as I start the next round of products I want to build.

If you have a product idea but cannot write the code, what would you build first—and what evidence would you require before anybody relied on it?

Reference information

This account draws on my February 2015 Foreign Fighter Detect Business Concept, the May–June 2024 xAlerts business case and Git records, ChatGPT and Claude exports covering 2023–2026, the AI Product Flow case studies, the Portfolio Dashboard acceptance records and relevant Post-Implementation Reviews. AI-generated summaries were used only when checked against the underlying files; they were not treated as my words.

subscribe to Our Resilient World

”When the world fundamentally changes we must respond with new approaches…”

We have passed a tipping point where the old way of doing business at the political level, the defence level, the environmental level, the societal level and potentially the economic level is not able to cope with the speed of change and more importantly the interactions between these changes. Something must change. Each of us is going to have to make a choice in the next few years - do we want to help design our future or do we have it done to us.

The ‘rule of law’ no longer applies

Humans find it easier to assume that the future will continue on a steady, upward path but any reading of history shows that the past is a series of phase transitions or ruptures that result in a ‘new normal’, a world that is fundamentally different from what has become before it.

President Putin and the return to the USSR

President Putin has been the most aggressive of the three great power Hegemons. As the table below shows Russia has been actively exercising its military power under his rule. There is every indication that Putin will not stop with his invasion of Ukraine unless he is defeated or removed.

The false premise that Putin promotes is that Ukraine belongs to Russia. The historical record as shown below shows just how false this premise is. Modern Russia has its origins in what is now Ukraine.

President Trump and the US break with ‘rules based order’

The US which acted as the world’s super power, was also largely responsible for what is called the ‘rules based order’. Most people suggest that this order ‘broke’ in 2025 with the re-election of President Trump and his aggressive tariff and military policies.

The new US Defense Strategy was released in January 2026.

This strategy exhibits a fundamental break from previous US Defense Strategy as shown in the table below. It also asserts its ‘might is right’ approach to the US domination of the Western World.

The US prior to President Trump exercised its military power leading coalitions it formed and organised. Under President Trump we have seen attacks on Venezuela and with Israel on Iran. These are the actions of a Hegemonic President and should not be seen as anything but an exercise in raw power. Combined with his use of tariffs as an exercise of economic power, President Trump has destroyed any belief in a ‘rules based order’ led by the country most responsible for creating and supporting this concept -at the same time as being one of the principle beneficiaries.

President Xi and return to ‘great China’

The third Hegemon that we need to fear is President Xi and the CCP. While China has wielded most of its power by economic means it has been active with military action.

The reason we need to fear President Xi and the CCP is that, just like Russia, China has argued that periods of its historical control give it the right to absorb other countries. The two most important examples are shown below:

Now that China has fully incorporated Tibet and gone a long way to destroying the Tibetan culture it is arguing that it has the right to control Taiwan, including by force if the Taiwanese people continue to want to remain independent.

The biggest lie by the CCP is it use of the nine-dash line to argue its right to control the seas within this zone. At no time has China had exercised administrative control of this zone, and this another example of how history is distorted to legitimize a grab for power.

China is not just using military force or the threat of military force to exert control. It actively uses trade coercion to demonstrate its hegemonic powers.

The new reality - US military is no longer preeminent

The war with Iran has demonstrated the limits of US power. In fact, Iran has won the war and we are just waiting to see what the end state will be. My assumptions is that Iran will control the Strait of Hormuz and will end up as a nuclear power. The only hope for the US to achieve any of its war aims is if the Iranian regime is overthrown.

The extent of this loss of military supremacy against a well prepared enemy is what is also happening in the Ukrainian/Russian war. In recent military exercises in which Ukrainian drone operators have participated they have played a dominant role and demonstrated that NATO military doctrine needed to be transformed.

“Exercise Hedgehog 2025 exposed an uncomfortable reality for NATO: the alliance’s logistics system, built for secure rear areas and deliberate decision cycles, cannot survive in a battlefield defined by constant surveillance, rapid targeting, and cheap attritable drones. In short, NATO’s system cannot be adapted, it must be transformed” (HEDGEHOG 2025: PRICKLY LOGISTICS IN A TRANSPARENT BATTLEFIELD)

What it means for each of us?

If you accept the premise that the world ‘rules based order’ for all its imperfections, is broken beyond repair, there are some hard choices to be made. I am influenced by my personal history of being actively involved in the anti-war movement in 1971 including anti-conscription siege at the University of Melbourne, and the Halt All Racial Tours campaign against the all white Springbok Tour. I remember shouting slogans like “Smash US Imperialism” at some of the big protests against the Vietnam War.

I also remember arguing with my father who had retired from the Australian Navy and who had joined the Navy as a naval cadet at 13, was at war by 17 and was involved in naval support during the Korean and Vietnam wars about protesting against the Vietnam War. I wasn’t clear if he supported the Vietnam war but he did think that the Americans had created an excuse to scale up its war effort in the Gulf of Tonkin incident. The one thing that my father did that has stuck with me every since was to ask me - “if you are against the war, what would you do instead?”. A question I am still struggling to answer.

Over time my opinions have changed - I am no longer ‘anti-war’ but I am also clear that Australians should not be sent to war overseas without Parliament voting to approve the war. One recommendation:

“For each military deployment, the Australian Government should provide and routinely update a clear statement of national interests and strategy. The strategy statement should include the elements of power to be used, the end state to be achieved, an indication of the exit strategy and likely time frame for the commitment of force.” (former Australian Army chief, Peter Leahy, quoted in ‘Parliament and the Australian way of war’, The Strategist, 2022

As a person who strongly advocates ‘participatory democracy’ I think we, the people, should be consulted to agree to the National Defence Strategy for each election cycle. As someone who is too old to fight, I don’t want, as a matter of principle, to ever commit anyone to a war. I also don’t want to see our armed forces committed to endless, futile wars like Vietnam, Iraq or Afghanistan as way to ‘buy US protection’.

Blame the Government, Not the People

The other thing we must do is to separate power structures from the people in these countries. I am fearful of the coercion that the CCP will use to exert control over the countries in its regional hegemony. I don’t blame, and I will never blame, the Chinese people or the Russian or Iranian people. They are controlled and brainwashed and are effectively powerless to resist - just look at the Tibetans, Hong Kong and Iranian people and see where resistance has got them.

Democratic Coalition - opt in only

I think people living in democracies where we do have the right and ability to change governments must stop trying to solve problems of ‘evil governments’. We need a new approach, one where we form a single super coalition of all the democratic countries. Leave the UN to do what it can but build a new coalition of willing participants to focus on things that benefit all these countries. Build resilience by cooperation with the only requirement that there are free elections overseen by a new independent electoral commission. This approach will be instantly opposed by the US who is already trying to break the EU as it doesn’t want another power bloc that it cannot coerce or bully.

Offer entry to this new coalition to any country that is prepared to agree to the principles and values of democratic countries and focus on defence, trade and resilience allowing each country to manage itself and its own coalitions and interests. Stop trying to fix broken countries with problems that cannot be solved externally and focus efforts on making a new coalition something the peoples in these countries want to join so that when an opportunity arises there is a clear path forward for them to choose to enter this democratic coalition.

Just like the EU there will be some conditions of entry around how a country qualifies, to ensure people have the basic freedoms necessary for a democracy to function and that corruption is not so widespread that it destroys effective democracy. Think of this democratic coalition as an “EU lite” - no single currency, no parliament, and limited bureaucracy, primarily to validate that elections are free and fair, with cooperation focused on defence and resilience.

Australia now needs to choose a strategy

It’s time we accept the reality that the US enforced ‘world order’ is broken and it will not be coming back. The US military defeat in its war with Iran has demonstrated that the US is no longer a military super power that can enforce its will anywhere in the world. The new ways of war, including the hybrid war being waged by Russia against Europe, combined with the inability of the US to protect its bases, results in a US that will not fight China over an Taiwan invasion.

If you accept this premise you need to rename Plan B to Plan A and start preparing for the new world of hegemons where middle powers like Australia cannot trust the US as much as we can’t trust China. The US only needs bases in Australia while they are prepared to try and contain China and protect Taiwan. Once the reality of their inability to fight China, close to China is accepted then Australia is irrelevant to US interests.

Australia needs partners - we need a NATO for the Pacific, where we agree with Japan, South Korea and any other countries who may be prepared to stand-up to China to sign a collective defence agreement. How long before Japan and South Korea look at developing their own nuclear deterrence when they can no longer rely on the US?

We need to follow Canada’s policy and reduce our economic dependencies on China as rapidly as possible. We need to support Ukraine and partner with them to develop new low-cost defence technologies as soon as possible, and most importantly we need Ukraine to win in its war with Russia. Australia has supported Ukraine. We need to fund and partner with leading Ukraine defence firms to licence the manufacture of all the major drone and missile categories in Australia while supplying these drones and missiles back to Ukraine. Canada has shown how much we should seek to contribute and we all need to demand more support for Ukraine.

The UK has shown what a joint partnership with Ukraine might look like. Developing this type of partnership with Ukraine and the UK would help reduce our reliance on the US as a defence supplier while accelerating our exposure to and use of the latest defence technologies. Ukraine’s focus on developing anti-ballistic missile systems in partnership with NATO members could also give us access to this technology at a much lower cost than we could obtain it from the US, and potentially within a few years instead of being near the back of the queue for access to Patriot anti-missile systems.

How participatory democracy could select a Defence strategy

If China were to try to take over Taiwan would people in Australia or New Zealand be asked if they would support the US and its allies in resisting this attack? If China does invade Taiwan the politicians in Australia and New Zealand won’t have time to ask if people support a war with China.

These type of decisions need to be discussed now. Taiwan and Ukraine have been excluded from any of these alliances because countries are afraid of the power that China and Russia can exert. Waiting too long guarantees that the hegemons including the US will use both economic and military coercion without effective push-back.

The chart below has been created as an example of what could be asked at the next election to help define the Government’s Defence strategy for the next electoral cycle. We can move from having to elect a government on its promises that is then capable of taking us into a war with no effective agreement from the Parliament, let alone the Australian people. I don’t think that this is good enough. Let me know in your comments or emails if you think that we should allow a government to take a country to war without any guidance at all from the people who will suffer and die in the war.

What about Israel and the Palestinians?

I need to be clear that I totally oppose the war Israel is waging on the Palestinian people. Israel has the right to exist but they are replicating what was done to them in the early stages of Hitler’s solution to the ‘Jewish problem’. At the same time I totally oppose the way Hamas has acted. Are they really fighting for the Palestine people or their own power over their people? This is a wicked problem and one that I don’t think anyone has a viable solution at this time.

Reference Information

Both AIs have checked the final version of these charts, maps and tables. Treat them with skepticism, but they provide information in minutes that would take me hours or longer to compile and create. This article has no AI input except for basic editorial review.

In my previous article, AI for a Resilient Australia: Who Sets the Priorities?, I suggested that the AI inquiries now under way may not ask the most important question: could AI help us move away from business-as-usual politics towards a different kind of politics, in which communities examine trade-offs and help set their own priorities?

Kate Chaney, the independent federal MP for Curtin, has published an AI Discussion Paper: Shaping the future of AI for Australia and invited feedback and ideas from everyone. I used my AI research workflow to prepare a detailed response: From Guardrails to National Resilience — Feedback on Shaping the Future of AI for Australia.

That paper was produced from my research and thinking, with OpenAI's Codex preparing the draft and Claude reviewing it. It argues that one practical demonstration could begin in Curtin:

Start with one issue nominated and framed by Curtin residents, not by an AI company, government department or parliamentary office. It could be a local resilience question involving housing, energy, care, transport, small business or the costs and benefits of digital infrastructure.

One of the best ways to understand what AI might make possible is to conduct real, small-scale experiments. This article outlines how an electorate such as Curtin could do that, building on tools and workflows I have prototyped.

What effective community consultation involves

I use consultation broadly: a structured process that enables people affected by a public question to understand it, contribute their knowledge and views, and influence the decision. A survey can be part of that process, but collecting opinions is not the same as giving people time, evidence and an opportunity to consider alternatives and trade-offs.

The OECD has done significant research into citizen participation, saying it ‘…should begin with a genuine problem the public can help solve. The expected influence of participants, who needs to be involved, and how the decision-maker will use and respond to their input should be clear from the start.’ AI Discussion Paper; OECD citizen-participation guidelines

My approach starts from the proposition that representative democracy and our BAU politics are not fit for purpose. Politics is a competition, and parties win not by seeking cooperation but by promoting division. The result is that we, the people are asked to vote once every three or four years and put up with the spectacle to the next electoral cycle.

Indeed it has been said that democracy is the worst form of Government except for all those other forms that have been tried from time to time.…’ Winston Churchill 2016

I think with the assistance of AI, it is time we start developing a better form of democracy and move from representative democracy towards participatory democracy.

A draft workflow for consultation assisted by AI

Ms Chaney has already used a community survey to understand her electorate's concerns about AI. If the survey captures a broad range of issues, AI tools could help organise those responses into a transparent model that lets people explore the relationships and trade-offs before providing further feedback.

To show how the idea might work, I built a demonstration for the Bega Valley, where I live.

Try the Bega Valley Community Resilience Simulation. It is a demonstration: nothing is saved and no personal information is collected.

Participants distribute 100 points among possible priorities. The constraint forces a choice: giving more weight to one issue leaves less for another.

My highest priority is preparation for the next severe bushfire season. That judgement is shaped by my experience of fires in the Bega Valley in 2018 and during the 2019–20 Black Summer fires.

The model also shows how issues that initially appear separate can overlap.

From priorities to testable options

The same method can be adapted to a specific issue. I created a Demo Bushfire Readiness Simulator to explore actions that might reduce risk before summer 2026–27.

I developed it in the context of the strong El Niño under way in August 2026. For my own situation, the model suggested that reducing hazards around my home was the most useful action I could take quickly. That is a result from my simulation using its declared assumptions. It is not general bushfire advice, a forecast, or evidence that the same intervention will be best for every household or community.

The research behind this demonstration was prepared using OpenAI's Codex and then reviewed with Claude. The sources and assumptions are exposed so that other people can question them.

Using a second AI model is a minimum check in my personal workflow. It is not enough but it does demonstrate what is possible. The AI policy research, policy papers and simulations are set up so that anyone can suggest a correction or provide new evidence.

Using established simulations to test our assumptions

I use the phrase citizen scientist when thinking about how people could find ways to cooperate together using these types of simulation tools. If people are to become more actively involved in setting priorities, they must be willing to adjust their opinions when better evidence or modelling challenges their assumptions. The most important test for anyone is not if your opinion is based on data, but are you prepared to change your opinions when you are exposed to new information. How easy is it to fail that test.

For example, I expected that rapidly reducing coal use would produce a much larger fall in projected global temperature by 2100. When I tested that idea in the EN-ROADS climate simulator, the result was less dramatic than I expected because changing one part of a connected energy system can affect other fuels and behaviours.

I have used EN-ROADS for more than five years, completed its online training, and reviewed the EN-ROADS Technical Reference.

I also asked Claude and ChatGPT to assess the documentation. The prompts were identical, with the addition of a request for ChatGPT to provide its results in a table format. The results are summarized below:

• Fundamental relationship between greenhouse-gas emissions, concentrations and global warming — Very high: Core physical relationships are well established.

• Broad behaviour of the global energy–climate system — High: Useful for learning how major parts of the system interact.

• Relative leverage of broad intervention packages — High: One of the model's strongest educational uses.

• Approximate global effect of a policy package — Medium–high: Results depend on socioeconomic, technology and implementation assumptions.

• Exact temperature from a slider combination — Medium: A scenario result, not a prediction.

• Economic damages — Low–medium: Estimates and modelling choices vary substantially.

• Outcomes for an individual country or community — Low: EN-ROADS is essentially a global model and is not designed for local forecasts.

• What governments and communities will actually do — Inappropriate use: The model explores “if this, then approximately that”; it does not predict human choices.

EN-ROADS is a sophisticated product that has been developed over many years but it is not a predictive model. What is does do though is provide a way for anyone who can wear the lens of ‘citizen scientist’ to examine the trade-offs and start to understand which choices have the biggest impact.

What a real electorate demonstration would require

The examples in this article are proofs of concept. They do not establish that participants are representative, that the results are accurate enough for a public decision, or that an MP will receive or act on them.

The aim is not to manufacture consensus. It is to help people understand the trade-offs between problems/opportunities. It is to provide our MPs with the opportunity to have a discussion around these trade-offs with their voters. It is to help people participate in government.

Conclusion

AI-assisted research and simulation create capabilities that were previously available mainly to governments, universities and well-funded organisations. They could now help ordinary people examine difficult choices—but only if the tools are transparent and allow anyone to challenge the assumptions and data and modelling aspects of the simulation.

I am continuing to test the approach. Another example is the Australian Bushfire Resilience Simulator.

The longer-term goal is to help people do more than vote once every three years and hope that the government they elect can move beyond business as usual. It is to create practical opportunities for citizens to work together, test hypotheses and change their views when the evidence warrants it.

I know that my own hypotheses and assumptions are often wrong. That is why I use AI models to challenge them and why I want the sources, assumptions and corrections to remain visible. The scientific habit that matters most is not simply relying on data. It is being willing to change our minds when the evidence shows that our original view was wrong.

If you have a testable hypothesis, add a comment. I will run suitable ideas through my AI Kangentic workflow and publish the result for others to question.

Subscribe to Our Resilient World

In the first story in this series, I argued that for the first time in history, humanity has tools that can help us cooperate at the same scale as the problems we need to solve.

If that is true, Australia needs to ask a much bigger question about artificial intelligence.

Most AI policy starts here:

• How do we regulate it?

• How do we manage the risks?

• How do we protect jobs?

• How do we control the building of data centres?

• How do we stop harmful uses?

All of those questions are important, but they are not enough. They start from the assumption that AI is mainly something being done to us, and that government's job is to control the consequences.

I want to start somewhere else.

How can Australia use AI to become more resilient by 2035?

My working definition of resilience is:

A resilient world is one in which people, communities and institutions have the capacity and ability to live well through disruption while continuing to shape a better future. It can meet essential needs, maintain social cohesion, sustain functioning economies and critical systems, protect the natural systems on which life depends, and adapt to technological, demographic, environmental and geopolitical change.— About Our Resilient World

A prediction I am prepared to be wrong about

Proposals for new AI policy have many starting points, and three are running in Australia right now. South Australia has announced a Royal Commission into Artificial Intelligence. Kate Chaney, the independent member for Curtin, has released an AI Discussion Paper asking for feedback on 18 policy proposals. The Australian Greens have started an inquiry into data centres.

These reports and policy proposals will contain many recommendations on how to control AI, and some will look at the opportunities AI creates. I predict (a testable hypothesis) that not one of them will make its central recommendation about how AI can be used to improve our democracy by empowering people to have more say in how decisions are made, or about how we can use AI to make Australia more resilient.

Looking at the current state of world and Australian politics, AI poses a very significant risk, because it can so easily be used to distort facts and promote division. That is exactly why the focus of AI policy should be on doing the opposite — using AI to empower people to take more control of their own lives, and to help them find common ground using the same approach that has been used to understand climate change, with one difference. This time, everyone can become a citizen scientist with the help of AI, provided they are given sufficient access to AI compute.

The South Australian Royal Commission will consist of a panel of experts who will hold hearings, take a great many submissions and put together a series of recommendations. That is the process we have always used, and we now have the option to do so much more than this.

Two ways to write AI policy. Both manage harm; only one changes who decides.

Start with the outcome

Before deciding what AI policy should contain, we need to decide what it is for.

Australia already has plenty of familiar policy goals: economic growth, higher productivity, greater competitiveness, more jobs, stronger national security. All of them are important, and none of them is sufficient, because a country can meet every one of those goals and still be more fragile at the end of it.

• A country can become more productive while becoming less cohesive.

• It can become wealthier while housing becomes less affordable.

• It can create new industries while the benefits flow mainly to a small number of people.

• It can build powerful digital infrastructure while the communities providing the land, energy, water and workers receive little long-term value.

That is why I prefer resilience as the goal. A more resilient Australia would be better able to anticipate change, protect what matters, recover from shocks and adapt as circumstances change. More capable, not just richer. More cohesive, not just more efficient. More able to make difficult choices openly, rather than waiting until each problem becomes a crisis.

If resilience is the goal, AI policy is no longer only about controlling a technology. It becomes a policy for expanding Australia's ability to learn, decide and act.

We should stop waiting to be asked

AI will not make Australia resilient by itself, and people remain responsible for the goals, the values and the final decisions. The opportunity is to give people far greater power to be involved and to shape their own future.

A resilient country is not one where the decisions are made by governments who have demonstrably failed the test of building the capacity for a resilient Australia. Business as Usual (BAU) government produces a BAU pathway to the future, in which nothing significant changes and each problem is dealt with only once it has become expensive enough to be unavoidable. All the really important problems we face are not separate problems but interacting systems — housing, energy, care, defence, climate and public revenue are one system, and no minister and no department owns it.

We must move to a world where ‘we the people’ have the ability to set the priorities for government. In Australia we need an AI policy whose most important question is how we use AI to give people the ability to tell governments what their priorities are, while understanding the trade-offs — that spending more on defence or health must mean spending less somewhere else.

Climate showed how far this can go, and where it stops

Climate science built two things that are worth copying.

The Intergovernmental Panel on Climate Change (IPCC) demonstrated that a distributed body of researchers can assemble evidence credible enough to move a global argument and to set out alternative pathways. The EN-ROADS climate simulator demonstrated something I think is more important still, which is that anyone — not only a modeller — can sit down and start to understand the trade-offs involved in reducing global heating.

EN-ROADS Climate Simulator— visualise the impact of policy choices and technology improvements.

Then comes the third step, and this is where it fails. The UN Conference of the Parties (COP) hands the decision back to government representatives, and the record shows we cannot rely on them to take effective action. Canadian Prime Minister Mark Carney spoke at Davos in January 2026 of a rupture in the world order, and said that we had always partly known the story of the international rules-based order to be false, because the strongest exempted themselves whenever it was convenient and trade rules were enforced asymmetrically.

The annual COPs just made this process unavoidably visible to all who cared to look. The evidence was never the bottleneck, and neither were the models. The decision step is the bottleneck, and thirty years of COP meetings have not moved it.

What climate science built, what the resilience equivalent would look like, and the step no international institution has been willing to take.

The first piece of this already works

We must develop the equivalent of the IPCC reports, driven by citizen scientists. We must develop a World Resilience Simulator that can drill down to democratic countries who are willing to act together on priorities that their citizens have discussed and then decided.

That is a very large ambition, so I built the smallest working piece of it to find out whether any of it is real. I took one question about AI and work out to 2035 — whether AI and robotics will mainly displace or reshape entry and re-entry work, or whether they can open viable business-creation pathways across age groups — and tested it across three reports and five separate hypotheses, using one AI agent to do the research and a second AI agent to check the first.

The Kangentic workflow — Codex (OpenAI) drafts and Claude checks Codex.

No commission, no secretariat and no terms of reference — and every verdict, limitation and source is public for anyone to challenge. Treat this as a demo of what can be done.

The answer came back negative. On the evidence available to 21 July 2026 the hopeful proposition is not supported. Business creation is real enough to investigate, but nobody can currently show that it is larger than displacement and augmentation, that targeted support would make it dominant, or that it would survive a major shock — largely because Australia does not collect the linked data on AI use, employment, business survival and income that would settle the question either way.

I want to be clear about why I am putting a negative result into an article arguing for optimism. A royal commission that spent two years and reported "we cannot yet answer this, and here is exactly what is missing" would be treated as a failure, which is precisely why commissions so rarely say it. In research that answer is the most useful thing you can produce, because it tells you what to measure next instead of what to announce next. The whole series is published with its verdicts, its confidence levels, its limitations and its sources, and with forms for anyone to submit a correction or new evidence that would change the result.

One person did that, with AI assistance, over a few weeks. The process is now automated and a new hypothesis (see AI Policy for more examples) can be tested and reviewed in a matter of hours. If a few thousand people do the same thing, publish their methods, label their evidence honestly and argue about the results in the open, then we have something no royal commission can produce: an evidence base that nobody owns, that anyone can audit, and that does not stop when the reporting date passes.

What people need in order to take part

Three things, and none of them is a degree in computer science.

The first is compute. Access to capable AI is now the difference between having an opinion about something and being able to test it. It should be treated the way we treat access to a public library, not as a consumer product that people are left to buy for themselves.

The second is AI literacy, and this should not mean teaching everyone to become a programmer. It should mean helping people understand what AI is good at, where it is unreliable, how to check its work, how to protect private information, and how to use it to extend rather than replace their own judgement. That belongs in lifelong learning, through schools, libraries, TAFEs, community organisations, workplaces and local government — not only through universities and large companies in capital cities. I will be writing more about this and demonstrating how AI can be used by anyone to do and share research in the role of a citizen scientist.

The third is time. A four day week has already been suggested as one way to reduce the potential loss of jobs caused by AI, but the more interesting benefit is that it gives people time to think — time to think about the world, about what is important to them, and about how they might be actively involved in shaping their own futures. Participation is not free, and we should stop pretending that it is.

Regional Australia should be first in line rather than an afterthought. Regional communities live with bushfire, flood, water, health, aged care, transport, agriculture, energy and small-business viability arriving as one connected problem. They are not managed that way. Councils, state agencies and federal programmes each hold one piece, on separate budgets, separate timetables and separate plans, and nobody is responsible for the interactions between them. That is precisely the gap that open models and citizen-run analysis could fill, because the people living with the connections are the ones best placed to describe them.

What is actually left for government

Less than the current debate assumes, and government is not doing that part well either.

Regulation and safeguards are genuinely necessary, and real harms need policing. Australia also needs enough sovereign capability to understand the systems it depends on, to protect sensitive information and to run critical services — informed interdependence rather than technological isolation, and a conscious choice rather than the accidental result of never building any capability of our own.

Data centres are the immediate test of whether any of this is understood. Data centres place real demands on land, electricity, water, construction capacity and network connections. Meeting the planning requirements is the floor, not the standard, and at present no approval process in this country asks the question that matters:

How will this infrastructure make the host community more resilient?

The expected costs, benefits and trade-offs should be modelled openly before approval and measured afterwards, and where the benefit cannot be shown it should not be assumed. On current arrangements it is assumed almost every time.

People remain responsible for the values and the trade-offs; AI makes those trade-offs visible enough to argue about honestly.

The choice

So Australians have a choice. We can treat AI mainly as a technology that needs to be controlled, which is the safe and familiar path and which ends in business as usual. Or we can treat AI as a capability to be used deliberately to build a more resilient country — building the evidence ourselves, modelling the trade-offs openly, and telling governments what our priorities are instead of waiting to be consulted about theirs.

To do this we must break our dependency on government to decide for us.

The question is not whether AI will change Australia. It will.

The question is whether we use that change mainly to protect existing systems, or to build something better.

Drafted by me, revised with AI.

Subscribe to Our Resilient World

AI can help people see connected systems and cooperate, while people remain responsible for values, goals and decisions.

Back in 2022 I wrote 2030 One World, a cookbook of ideas for how people could work together to change the world. The central idea was that a significant number of people had stopped waiting for governments, political parties and powerful organisations to decide the future for them, and had begun taking greater responsibility for shaping it themselves.

Not as isolated individuals, and not by trying to force everyone else to agree with them. They did it by choosing to cooperate. The approach was voluntary: people and communities opted in because they accepted some shared principles and believed they could achieve more together than they could separately.

AI had a role in that vision, although the technology available at the time was nowhere near what we have now. Its role was not to govern people or decide what was right. It was to help people clarify their ideas, examine evidence, compare different futures and work backwards from the future they wanted to build.

At the time I could imagine the destination more easily than I could explain how we might get there. That is beginning to change.

What has changed

The public debate about artificial intelligence is dominated by one question, and it is the wrong one to build on:

But there is another question I want to explore:

Over the past few years I have used AI to do work that would previously have required teams of researchers, analysts, designers and programmers – or would simply have remained beyond my reach.

The tools are far from perfect. They make mistakes, they can sound confident when they are wrong, and they reflect the limitations of their data and design. But in my experience their usefulness has grown at a remarkable rate. They can already help an ordinary person examine a complex subject, organise large amounts of information, compare competing arguments, identify assumptions, draft models, explain difficult ideas and turn a vague intention into a practical plan.

Each of those capabilities matters on its own. Together, they change what ordinary people are able to do.

The problem is that everything is connected

Most of the serious problems facing us do not sit neatly inside one government department, one community or one country. Housing connects to migration, planning, finance and employment. An ageing population connects to health, care, workforce participation and productivity. Climate disruption connects to infrastructure, insurance, migration and social cohesion. Artificial intelligence connects to employment, productivity, energy use and the distribution of power.

The numbers show how many of the four named problems touch each shared domain.

These are not separate problems. They are interacting systems. Yet our institutions still divide them into separate portfolios, reports, budgets and political arguments.

That is understandable — human beings have limited time and attention, so we simplify complicated issues in order to deal with them. The problem is that the world does not stay simplified just because our institutions are organised that way. A decision that improves one part of the system can create new problems elsewhere. A policy that looks effective over three years can make the country less resilient over ten. A national solution can fail when the most important drivers operate across borders.

We need help seeing more of the system. AI may provide some of that help.

What AI does, and what stays with us

I do not think AI should decide the future for us. Only people can decide what kind of society they want to live in, what trade-offs they are prepared to accept and what they owe to one another.

People decide:

• Values and priorities

• What trade-offs are acceptable

• What we owe one another

• The final call

• Who is accountable

AI can help by:

• Making assumptions visible

• Bringing together evidence from different fields

• Showing where experts agree and disagree

• Exploring the consequences of different choices

• Explaining complex models in language more people can understand

• Recording decisions and unresolved questions so groups can learn from earlier attempts

Most importantly, it makes these capabilities available beyond governments, universities and large corporations. That is the shift. We now have tools that could help us cooperate at the same scale as the problems we need to solve.

The word could is doing real work in that sentence. AI will not create cooperation by itself, and it can just as easily be used to manipulate, divide and concentrate power. The opportunity exists only if people choose to use these tools differently.

Why resilience, and why 2035

Every serious policy discussion should begin with a goal: what are we trying to achieve, what does success look like, and how will we know whether we are moving towards it? Too often the debate starts with a policy answer before there is any agreement about the outcome. One side proposes more government spending, another proposes lower taxes; one argues for regulation, another for competition. It becomes a contest between positions, with the goal assumed rather than stated.

I want to start with a simpler goal: a more resilient world by 2035. Not a perfect world, and not a world without conflict, inequality or risk — a world better able to anticipate shocks, preserve what matters, recover when systems fail and adapt when circumstances change.

Resilience gives us a direction without pretending everyone will agree on every detail, and it forces us to consider the whole system:

• A country is not resilient if its prosperity depends on fragile supply chains it cannot influence.

• A community is not resilient if the people who keep it functioning cannot afford to live there.

• An economy is not resilient if productivity gains flow mainly to those who already hold wealth and power.

• A democracy is not resilient if citizens cannot distinguish evidence from manipulation, or see how major decisions were reached.

• A world is not resilient if every country protects itself while the global systems they all depend on keep weakening.

2035 is close enough to matter and far enough away to make structural change possible, without becoming an excuse for postponement. Most of us struggle to imagine 2050 in practical terms, and it is easy for governments to announce distant targets that someone else will have to deliver. The decisions made in the next few years will shape what is possible by 2035 — and AI's capabilities will change dramatically over that period even if the current rate of improvement slows. We do not need to predict exactly what the technology will become. We need to decide how we want to use the capabilities already in front of us.

Personal responsibility does not mean acting alone

"Personal responsibility" is usually code for the idea that individuals should simply look after themselves. That is not what I mean. No individual can solve climate disruption, redesign democracy, rebalance the economy or manage technological change.

Personal responsibility begins with deciding not to remain only a spectator. It means accepting some responsibility for the systems we participate in and the future those systems are creating. But responsibility only becomes useful when people can cooperate.

The scale of the problems means cooperation must eventually work across borders. I do not have a settled institutional design for that cooperation, and I do not want to assume an answer at the beginning of the series. The destination is a more resilient world built through voluntary cooperation. How that cooperation should be governed is one of the questions the series must examine openly.

That does not begin with a new institution being declared into existence. It begins with the people already trying to help and most importantly it begins with the people who care but can’t see anyway they can make a difference. Like me they have held of on getting involved because they see that the problems we have to solve effect the world, not their nation or community. These problems can’t be solved by focusing on climate change, inequality, etc etc. These problems can only be solved by voluntary collaboration to define a future state and then working backwards to decide how to move towards it.

Many of us understand the problems facing the world need to be thought about as ‘world’ problems. The opportunity AI gives us is the ability for individuals who have felt powerless to start to find ways to cooperate with others to see if they can define what a 2035 resilient world could look like and then start to work to create it. My task and my commitment is to stop thinking about how we might do this and to start to build examples of how we can use AI to do things that just have not been possible before.

Voluntary, or it is not worth building

This is why voluntary participation matters. Any system built around these ideas has to be voluntary. It cannot be another attempt to impose a single answer on everyone. People, organisations and countries will disagree about values, priorities and acceptable trade-offs, and some will not want to participate at all. That is their choice. The aim is not universal agreement — it is to create better ways for those who do want to cooperate to do so openly, fairly and effectively, at the scale the problems actually operate on.

That means the assumptions must be visible, the evidence open to challenge, alternative pathways genuinely considered, models improved when reality shows they are wrong, and decisions accountable to the people affected by them.

AI could make this kind of participation far more practical, but it cannot guarantee it. The design of the institutions still matters, and those questions are human ones:

• Who owns the systems?

• Who can inspect them?

• Who can contribute?

• Who benefits?

• Who can challenge a decision?

An exploration, not a finished answer

Our Resilient World will work through these ideas over time. I do not have a finished blueprint, and I do not believe any individual, government, company or AI model could produce one — the challenge is too complex and the future will keep changing.

What we can do is set a direction, explore different pathways, test small parts of the approach and learn openly from the results. Over the coming months I want to ask what a more resilient Australia might look like by 2035, why that resilience depends on cooperation beyond our borders, and what an opportunity-first AI policy would actually look like. I want to explore how open models, simulations and backcasting can help people move from a preferred future to practical action — and build small prototypes, not because they are the answer, but because working examples let ideas be tested.

That is where I want to begin. Not with another argument about whether AI is good or bad, and not with a promise that technology will save us. With a more practical question:

Can we use these new tools to help ordinary people understand complex systems, cooperate more effectively and take a greater role in building the future?

I think the opportunity is real. Whether we use it is up to us.