AI Inquiry Atlas

An unofficial, independent guide to the public records of the Joint Select Committee on Artificial Intelligence. Not connected with the Parliament of Australia or the committee. Its makers also made a submission (no. 481; see About). Summaries and scores are written by AI and can be wrong: check the original before relying on them.

Submission 277

Monash University

Submitter group: Universities.

An AI model identified the submitter group and wrote the summary, key points and recommendation summaries. Counts do not measure importance or merit.

Open in the interactive guide

Summary

Read the original submission for the full text.

Argues that Australia should pursue an ambitious, outcomes-led national AI agenda built on sovereign capability, including computing infrastructure, research, workforce pathways and Indigenous participation. Addresses each term of reference, favouring a gap analysis with targeted reform, a Commonwealth assurance framework and strategic-resilience measures.

Key points

Quotes are the submitter’s own words. Text taken from documents, including scanned pages read by text-recognition software, can differ from the PDF. Check the original. Listed in the same order as on the submission’s page in the interactive guide.

  1. Monash argues that sovereign capability covers people, institutions, infrastructure, data and research capability, not only domestic frontier models.

    Sovereign capability is not solely the ability to develop frontier AI models domestically. It also comprises the people, institutions, infrastructure, data assets and research capability to develop, adopt, evaluate and govern AI technologies.

  2. Says substantial contraction of internships, graduate programs and entry-level roles risks weakening the pipeline for future expertise, professional capacity and leadership.

    If internships, graduate programs and entry-level roles contract substantially, the consequences will extend beyond the immediate prospects of young people. Australia risks weakening the pipeline through which future expertise, professional capacity and organisational leadership are developed.

  3. Monash argues that regulatory reform should begin with a systematic gap analysis of particular AI uses and harms.

    Reform should begin with a systematic gap analysis, examining particular uses of AI, the harms that may arise, and the extent to which existing legal, regulatory and assurance mechanisms address them.

  4. Monash argues that responses to AI-enabled harm should address the broader technological and commercial ecosystems, not only individual perpetrators.

    Effective responses should include a focus on individual perpetrators, but also extend to address the broader technological and commercial ecosystems that enable AI-facilitated harm.

  5. Monash argues that the Commonwealth should exemplify responsible AI adoption, with accountability remaining with clearly identified officials, supported by impact assessment, documentation, auditability and monitoring.

    The Commonwealth should be an exemplar of responsible AI adoption. Accountability should remain with clearly identified officials, supported by appropriate impact assessment, documentation, procurement assurance, auditability, security and continuous monitoring.

  6. Monash argues that independent evaluation capability should be treated as strategic national infrastructure.

    Evaluation capability should be regarded as strategic national infrastructure.

Recommendations

Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.

  1. Adopt a coordinated, outcomes-led national AI strategy that aligns major government funding and policy initiatives and supports safe, human-centred deployment.

    a coordinated, outcomes-led national AI strategy that aligns major government funding and policy initiatives and supports safe, human-centred deployment

  2. Integrate planning of computing and electricity infrastructure.

    integrated planning of computing and electricity infrastructure

  3. Ensure national AI investment and priority setting creates pathways for Indigenous-led research, innovation and participation in Australia's AI capability.

    ensuring national AI investment and priority setting creates pathways for Indigenous-led research, innovation and participation in Australia's AI capability.

  4. Make sustained national investment in sovereign AI capability covering people, research, computing infrastructure, trusted data environments, independent evaluation, Indigenous participation and secure access to advanced global technologies.

    a sustained national investment in sovereign AI capability encompassing people, research, computing infrastructure, trusted data environments, independent evaluation, Indigenous participation and secure access to advanced global technologies.

  5. Coordinate and co-design Indigenous AI workforce pathways with higher education, VET, key industry groups and employers, extending participation to technical, research, governance and leadership roles.

    This workforce agenda should include deliberate pathways for Aboriginal and Torres Strait Islander people into AI-related education, research and employment. Indigenous participation should extend beyond AI literacy and access to technology to technical, research, governance, and leadership roles with Australia's AI ecosystem. This requires coordination and co-design with higher education, vocational education and training, key industry groups and employers.

  6. Develop work-integrated learning, paid placements, apprenticeships and supported early-career employment so AI augments human capability without removing its foundations.

    New models of work-integrated learning, paid placements, apprenticeships and supported early-career employment will be needed to ensure that AI augments the development of human capability rather than inadvertently removing its foundations.

  7. Address immediate skills needs while building adaptive capability and lifelong learning pathways for ongoing technological change.

    Policy should address immediate skills needs while building the adaptive capability and lifelong learning pathways that enable Australians to respond to ongoing technological change.

  8. Monitor the immediate effects of AI on job creation and displacement and act in a coordinated way to redesign and preserve pathways into professional work, including paid placements and work-integrated learning.

    monitoring the immediate effects of AI on job creation and displacement, and coordinated action to redesign and preserve pathways into professional work, including through paid placements and work-integrated learning

  9. Develop Indigenous AI workforce and research pathways spanning education, higher degree research, early-career development and research leadership.

    developing Indigenous AI workforce and research pathways spanning education, higher degree research, early-career development and research leadership.

  10. Support internationally interoperable transparency and metadata standards that let users understand the provenance and operation of AI systems.

    supporting internationally interoperable transparency and metadata standards that enable users to understand the provenance and operation of AI systems

  11. Invest in developing and evaluating open-access, responsible and energy-efficient AI tools.

    investing in the development and evaluation of open-access, responsible and energy-efficient AI tools.

  12. Invest across the full discovery-to-impact continuum, including mission-led research, testbeds, industry partnerships and commercialisation pathways.

    investing across the full discovery-to-impact continuum, including mission-led research, testbeds, industry partnerships and commercialisation pathways

  13. Develop stronger national pathways for AI research commercialisation, including support for scientist entrepreneurs, deep-technology spinouts and retention of high-growth start-ups in Australia.

    developing stronger national pathways for AI research commercialisation, including support for scientist entrepreneurs, deep-technology spinouts and the retention of high-growth start-ups in Australia

  14. Where Indigenous knowledge, data or community participation contributes to research or commercial value, establish appropriate governance and benefit-sharing arrangements from the outset.

    Where Indigenous knowledge, data or community participation contributes to research or commercial value, appropriate governance and benefit-sharing arrangements should be established from the outset.

  15. Support Indigenous-led AI research, innovation and commercialisation, including mechanisms for governance, attribution and benefit-sharing.

    supporting Indigenous-led AI research, innovation and commercialisation, including appropriate mechanisms for governance, attribution and benefit-sharing.

  16. Support AI inclusion through place-based partnerships, workforce development, shared infrastructure and community co-design.

    AI inclusion should be supported through place-based partnerships, workforce development, shared infrastructure and community co-design.

  17. Take place-based approaches to AI adoption that align low-latency digital infrastructure, workforce development and shared capability with local priorities, embedding First Nations leadership, knowledge and data governance.

    place-based approaches to AI adoption that align low-latency digital infrastructure, workforce development and shared capability with local priorities, while embedding First Nations leadership, knowledge and data governance.

  18. Limit tools designed primarily for abusive purposes.

    limitations on tools designed primarily for abusive purposes

  19. Inform policy with multidisciplinary evidence and lived experience, using technology-neutral, harm-based regulatory frameworks that remain effective as AI capabilities evolve.

    Policy should be informed by multidisciplinary evidence and lived experience, and regulatory frameworks should be technology-neutral and harm-based to ensure they remain effective as AI capabilities continue to evolve.

  20. Address harms from over-reliance in risk frameworks and design effective oversight around task/system design and people's capacity and incentive to intervene.

    Risk frameworks must also address harms that can arise when AI operates as intended but is relied upon too heavily. Effective human oversight depends on how the task and system are designed, and whether the person retains the capacity and incentive to intervene.

  21. Use AI risk frameworks that distinguish deliberate misuse from harms arising through over-reliance and convergence of professional judgement.

    AI risk frameworks that distinguish deliberate misuse from harms arising through over-reliance and convergence of professional judgement

  22. Respond to AI-enabled harm with safety-by-design, platform responsibility, provenance, detection, education, accessible redress and rapid incident response, prioritising children and vulnerable communities.

    safety-by-design, platform responsibility, provenance, detection, education, accessible redress and rapid incident response, with priority given to children and vulnerable communities.

  23. Conduct a nationally coherent, risk- and rights-based review of existing laws by AI use case and potential harm, with targeted reform where gaps cannot be addressed through existing frameworks.

    a nationally coherent, risk- and rights-based review of existing laws by AI use case and potential harm, with targeted reform where demonstrated gaps cannot be addressed through existing frameworks

  24. Allocate responsibility according to system capabilities, deployment context and each party's degree of control when assessing legal gaps.

    Any assessment of legal gaps should therefore consider how responsibility should be allocated according to the system’s capabilities, context of deployment and the degree of control exercised by each party.

  25. Keep Australia's approach nationally coherent, interoperable with major international partners and supported by clear regulatory responsibilities.

    Australia’s approach should be nationally coherent, interoperable with those of major international partners, and supported by clear regulatory responsibilities.

  26. Recognise collective AI-affected interests, particularly Indigenous data, knowledge, cultural material and communities, where individual consent or privacy may inadequately address collective rights and governance.

    Regulatory frameworks should also recognise that some interests affected by AI are collective rather than solely individual. This is particularly relevant where AI systems use or affect Indigenous data, knowledge, cultural material or communities, where conventional approaches centred on individual consent or privacy may not adequately address collective rights and governance interests.

  27. Allocate regulatory and private-law responsibility clearly, supported by coordinated assurance mechanisms, practical guidance and regulatory test environments.

    clear allocation of regulatory and private-law responsibility, supported by coordinated assurance mechanisms, practical guidance and appropriate regulatory test environments.

  28. Invest in governed Australian datasets, secure infrastructure, benchmarks and domain-specific AI models combining Australian data with sectoral expertise and independent evaluation.

    investment in governed Australian datasets, secure infrastructure, benchmarks and domain-specific AI models that combine Australian data with sectoral expertise and independent evaluation

  29. Establish data-governance arrangements protecting privacy, security, consent, ownership and benefit-sharing, embedding First Nations leadership and Indigenous data sovereignty.

    data-governance arrangements that protect privacy, security, consent, ownership and benefit-sharing, and embed First Nations leadership and Indigenous data sovereignty.

  30. Give Indigenous peoples authority over collection, access, interpretation, linkage, AI training, commercialisation and reuse of Indigenous data, including deciding whether particular uses should occur.

    Indigenous Data Sovereignty should encompass authority over how Indigenous data is collected, accessed, interpreted, linked, used to train AI systems, commercialised and reused, including the ability to determine whether certain uses should occur at all.

  31. Retain accountability with clearly identified Commonwealth officials, supported by impact assessment, documentation, procurement assurance, auditability, security and continuous monitoring.

    The Commonwealth should be an exemplar of responsible AI adoption. Accountability should remain with clearly identified officials, supported by appropriate impact assessment, documentation, procurement assurance, auditability, security and continuous monitoring.

  32. Enable people affected by decisions to understand when and how AI was used, challenge decisions and access timely human review and redress.

    People affected by decisions should be able to understand when and how AI has been used, challenge relevant decisions and access timely human review and redress.

  33. Require graduated testing before broader deployment of higher-risk AI, clear governance and particular consideration of children, vulnerable Australians and communities facing disproportionate consequences.

    Higher-risk uses require graduated testing before broader deployment, clear governance arrangements and particular consideration of children, vulnerable Australians and other communities that may experience disproportionate consequences.

  34. For Commonwealth AI materially affecting Aboriginal and Torres Strait Islander peoples or communities, include Indigenous participation in design, impact assessment, governance and evaluation.

    Where Commonwealth AI systems materially affect Aboriginal and Torres Strait Islander peoples or communities, Indigenous participation should form part of system design, impact assessment, governance and evaluation.

  35. Have departments and agencies evaluate AI initiatives and, where appropriate, publish findings about their performance, limitations and impacts.

    Departments and agencies should evaluate AI initiatives and, where appropriate, publish findings about their performance, limitations and impacts.

  36. Adopt a consistent Commonwealth assurance framework for AI adoption requiring public value, accountable officials, proportionate impact assessment, graduated testing, transparency, auditability, human review, contestability, redress and monitoring.

    a consistent Commonwealth assurance framework for AI adoption requiring clear public value, accountable officials, proportionate impact assessment, graduated testing, transparency, auditability, human review, contestability, redress and continuous monitoring

  37. Require appropriate community participation and assessment of differential impacts where high-impact government AI systems affect Aboriginal and Torres Strait Islander peoples.

    requiring appropriate community participation and assessment of differential impacts where high-impact government AI systems affect Aboriginal and Torres Strait Islander peoples.

  38. Give Aboriginal and Torres Strait Islander peoples authority over use of Indigenous cultural material and knowledge in AI training, model development and commercial applications, with mechanisms for consent, attribution, governance and benefit-sharing.

    Aboriginal and Torres Strait Islander peoples should have authority over the use of Indigenous cultural material and knowledge in AI training, model development and commercial applications, with mechanisms for consent, attribution, governance and benefit-sharing.

  39. Consider how existing IP frameworks apply to AI-generated and AI-assisted creative work, with particular protection for Indigenous cultural and intellectual property.

    careful consideration of how existing intellectual property frameworks apply to AI-generated and AI-assisted creative work, with particular protection for Indigenous cultural and intellectual property.

  40. Assess resilience against cyber compromise, loss of power or communications, externally provided model changes, automated-control failures and simultaneous behaviour of large AI-related loads.

    Resilience assessments should consider cyber compromise, loss of power or communications, changes to externally provided models, failures of automated controls and the simultaneous behaviour of large AI-related loads.

  41. In high-consequence settings, support governed decision-making, diagnostics and scenario analysis with independent validation, accountable human oversight and safe fallback arrangements.

    In high-consequence settings, AI should support governed decision-making, diagnostics and scenario analysis, with independent validation, accountable human oversight and safe fallback arrangements.

  42. Invest in secure AI and cyber research capability, supply-chain assurance and resilience of the critical infrastructure on which AI depends.

    sustained investment in secure AI and cyber research capability, supply-chain assurance and the resilience of the critical infrastructure on which AI depends

  43. Adopt adaptive, risk-based safeguards protecting genuinely sensitive technologies while maintaining trusted international research collaboration and secure data-sharing.

    adaptive, risk-based safeguards that protect genuinely sensitive technologies while maintaining trusted international research collaboration and secure data-sharing arrangements.

  44. Review and adapt the AI Safety Institute's remit and capabilities as the risk environment evolves.

    While the Australian AI Safety Institute’s initial focus may appropriately emphasise technical and cyber risks, AI safety also encompasses a broader and interconnected range of economic, civic, social, health and child-safety risks. Its remit and capabilities should be reviewed and adapted as the risk environment evolves.

  45. Treat national AI evaluation capability as strategic infrastructure, with the AI Safety Institute drawing systematically on multidisciplinary Australian research expertise for continuous testing, horizon scanning and emerging-risk assessment.

    treating national AI evaluation capability as strategic infrastructure, with the Australian AI Safety Institute drawing systematically on multidisciplinary Australian research expertise for continuous testing, horizon scanning and emerging-risk assessment

  46. Assess dependence on external AI providers as a strategic resilience risk, including mapping systemic dependencies and establishing continuity and contingency arrangements.

    assessing dependence on external AI providers as a strategic resilience risk, including mapping systemic dependencies and establishing continuity and contingency arrangements.

Read the original

Open in the interactive guide