# Federal AI Inquiry — submission workbench Status: research and draft position. Not lodged. Canonical policy outline: `AI_POLICY_POSITION.md`. This workbench maps and develops that position for this specific inquiry. ## Official inquiry - Joint Select Committee on Artificial Intelligence, Parliament of Australia. - Appointed by the House of Representatives and Senate on 20 August 2026. - Submissions close: 14 September 2026. - Reporting date: 30 November 2026. - Contact: ai.joint@aph.gov.au; +61 2 6277 3863. - Official page: https://www.aph.gov.au/Parliamentary_Business/Committees/Joint/Artificial_Intelligence/ArtificialIntelligence - Published submissions: https://www.aph.gov.au/Parliamentary_Business/Committees/Joint/Artificial_Intelligence/ArtificialIntelligence/Submissions - Monitoring status on 28 August 2026: the official submissions list reports no records to display. - Submission guide: https://www.aph.gov.au/Parliamentary_Business/Committees/Senate/How_to_make_a_submission ## Scope The terms of reference cover AI opportunities, productivity and living standards; regional and suburban distribution of benefits; AI and robotics; adoption barriers for small, medium and family business; sovereign capability and Australian data; workforce and skills; Commonwealth services; laws and regulation; copyright; harms; cyber and national security; emerging AI capability; and related matters. ## Main submission position AI and robotics are coming like a freight train. They cannot realistically be paused or stopped. The government’s task is to guide a transition already gathering momentum, like changing the course of a huge ship, and to help people understand why the outcome can be positive. This should be an optimistic, human-centred submission. It should lead with the extraordinary benefits that may be created when intelligence, robotics, energy and production become more capable: less dangerous and repetitive work, better care, wider access to expertise, stronger regional and small-business capability, and more time for chosen human activity. Security and risk should be treated as necessary supporting safeguards, not the main story. The people investing enormous sums and repeatedly saying that they want AI to improve life for everybody may be viewed with a degree of caution, but their investment shows that the builders and funders believe the technology can deliver. Government should acknowledge that confidence while making public benefit, honest explanation and accountability real. ## Core points from Martin — 28 August 2026 ### Explain the physical infrastructure plainly The government must address public concern about data centres directly and early, not leave a basic legitimacy problem to be filled by slogans or speculation. It should explain, in ordinary language: - what data centres are and what they do; - how their benefits are intended to reach everybody, not only technology companies or large institutions; - where the energy, land and water impacts occur and who bears them; and - exactly how water used in data-centre operations is managed, replenished, recycled or otherwise prevented from becoming a destructive drain on shared water resources. This should be a standing public explanation with evidence, local figures, assumptions and accountability—not a one-off announcement after controversy has already hardened. ### Keep capability open where possible The submission should address open-weight AI as part of the question of who gets to participate in the future. Open weights may support local experimentation, research, competition, education and Australian capability, but the submission should distinguish openness of model weights from responsible safeguards, secure deployment, data protection and accountability. The final wording should be tested against the committee’s sovereign-capability, data-sovereignty, adoption and safety terms of reference. ### Education must make the future imaginable and explain the transition Public education should do more than warn people about risks or teach them to operate a product. The government needs to work hard to explain the tremendous benefits that may follow from AI and robotics, paint a credible picture of what ordinary life could become, and explain the path from the present system to that future. That explanation must answer the basic economic question: if AI and robots do much more of the work, how do people live and get paid? A serious public account should discuss possible transition mechanisms—universal basic services, citizen dividends, income or productive assets owned through personal agents, and the possibility that lower costs reduce the amount of income people need—without pretending that any one mechanism has already been chosen. The central issue is distribution: who owns the productive systems, who can use them, and how people retain agency and dignity while work changes. This is not an argument for government propaganda. It is an argument for government to explain both the opportunity and the mechanism, so people can judge a real future rather than react to an unexplained infrastructure build-out. ### Make the transition emotionally and intellectually intelligible The government needs to reassure people without patronising them. The fear that accompanies a major technological change is understandable, but reassurance should come from a credible picture of what is changing, what is likely to improve, what remains uncertain and how citizens share in the result. The New Earth Catalogue is part of this work. Its purpose is to paint a realistic possible world, including the infrastructure hidden behind it, and to ask what government will need to provide in a post-agentic society: energy, water, food, housing, mobility, education, health, care, income or universal services. This turns optimism into something people can examine rather than a slogan. The submission should also ask a deeper question about increasingly capable systems: is the mind in an AI us, a version of us or a separate kind of mind? The practical phrase is: **who’s minding the robot, and what’s minding the technology?** It should remain a visible question about responsibility as systems become more capable and embodied. ### Build the people and institutions that make rights usable The government should employ and train more lawmakers and the wider legal-policy workforce needed to keep pace with AI: legislative drafters, lawyers, legal researchers, regulators, technical experts and public-interest advisers. This is not only about writing new laws. It is about improving access to the law, timely interpretation, enforcement, review and appeal. A right that exists on paper but cannot be understood or reached by an ordinary person is not an effective safeguard. ## Evidence to gather 1. Martin’s direct experience building practical AI systems, agents, websites and public-facing experiments. 2. The difference between access to AI tools and meaningful capability: literacy, local support, data, cost and confidence. 3. Examples of AI improving care, small-business productivity, public services and regional participation. 4. Risks and failure modes: opaque decisions, scams, deepfakes, data misuse, exclusion and concentration of power. 5. Specific recommendations mapped to the committee’s terms of reference. 6. Evidence about data-centre water use, cooling methods, local impacts, replenishment/reuse and public reporting. 7. Evidence and examples concerning open-weight models, local capability, research access and responsible release. 8. Evidence about legal capacity, legislative drafting, regulatory staffing, public-interest advice and practical access to review and appeal. 9. Evidence about the public benefits and transition mechanisms that can make an optimistic post-agentic future credible. ## Requirements from the Parliament submission guide - Address some or all of the terms of reference; there is no need to cover every item. - Keep the submission concise, generally no longer than four to five pages. - Begin with a short introduction to the person or organisation making the submission and make the key points easy to see. - Explain both the problem and how it should be addressed; the committee is looking for ideas that can become recommendations. - Include only supporting documents that directly relate to the key points and material that could responsibly be published online. - Provide return contact details with the submission, but keep personal contact details out of the main substantive text where possible. - If confidentiality is requested, say so clearly on the front and explain why. - Martin remains responsible for the accuracy of the submission, including any material developed with AI-assisted writing tools. - The research workbench can be extensive, but the final submission should normally be a concise 4–5 page document with a short summary and a small number of well-supported recommendations. - Preserve the longer policy position separately, then select the strongest supported recommendations for the final filing. ## Filing checklist - Keep the final document unpublished until lodged, because Parliament’s guidance says submissions should be prepared solely for the inquiry and not published elsewhere. - Put a short summary and numbered pages at the front. - Include return email/postal address and contact number, but keep personal details out of the substantive body where possible. - Mark any confidential material clearly and explain the request. - Lodge through the official inquiry page using a My Parliament account, then retain the acknowledgement email or receipt. This workbench and the public page are not evidence of lodgement or parliamentary acceptance. ## Monitoring plan - Use the inquiry page’s `Track Inquiry` control for official updates, subject to confirmation before subscription. - Recheck the official published-submissions page as the inquiry accepts material. - For each published submission, record submitter, date/number, terms-of-reference themes, useful evidence, recommendations, and points of agreement or difference.