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 46

PwC

Submitter group: Companies.

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 AI's economic value depends on moving from pilots to scaled adoption, supported by infrastructure and sovereign capability, workforce investment and trust. Notes public caution about AI risks. States that a standalone AI Act is not the right starting point and recommends closing targeted gaps in existing laws, regulator coordination and assurance.

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. Says around two-thirds of medium to large Australian businesses report some AI use, but fewer than one in ten describe adoption as significant or at scale.

    Yet the gap between experimentation and transformation remains significant. Around two-thirds of medium to large Australian businesses report some AI use, fewer than one in ten describe that adoption as significant or at scale.

  2. Says sovereign AI capability extends beyond infrastructure to research, talent, data, intellectual property and commercial capability.

    Sovereign capability also extends beyond physical infrastructure to research, talent, data, intellectual property and commercial capability required to develop and retain value in Australia.

  3. Argues that policy should protect entry-level pathways, enable redeployment and continuous reskilling, and support groups disproportionately exposed to disruption.

    The policy task is therefore to manage the composition of the transition: protecting entry-level pathways, enabling redeployment and continuous reskilling, and ensuring that groups disproportionately exposed to disruption are not left behind.

  4. Argues that existing laws already address many AI risks and that the issue is that they were largely designed around human decision-making.

    Australia’s existing laws already address many of the risks associated with AI. Privacy, consumer, corporations, anti-discrimination, work health and safety, and prudential laws generally apply regardless of the technology being used. The central issue is not an absence of regulation, but that these laws were largely designed around human decision-making.

  5. Suggests addressing specific gaps in existing laws and coordinating regulators, saying a standalone AI Act is not the appropriate starting point.

    Consistent with a technology-neutral approach, we do not consider a standalone AI Act to be the appropriate starting point. A more practical approach might be to address specific gaps in existing laws, coordinate action across regulators and clarify what organisations must do, and demonstrate, to comply.

Recommendations

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

  1. Assess AI's economic contribution by the value it creates across the economy (revenue, productivity, margins, capacity, quality, competitive position), not by how widely it is used or how many use cases are deployed.

    The economic contribution of AI should be assessed by reference to the value it creates across the economy (including higher revenue, greater productivity, improved margins, increased capacity, better quality or a stronger competitive position), not simply by how widely it is used or how many use cases are deployed.

  2. Explore how national measurement frameworks, such as the National AI Centre's AI Adoption Tracker, could better measure the depth and impact of AI adoption, including whether anticipated business outcomes are realised.

    Explore how national measurement frameworks, such as the National AI Centre’s AI Adoption Tracker, could better measure the depth and impact of AI adoption, including the extent to which anticipated business outcomes are realised.

  3. Expand practical support for small businesses to access shared AI tools, industry platforms, trusted services and implementation support.

    Government could expand existing practical support for small businesses to access shared AI tools, industry platforms, trusted services and practical implementation support.

  4. Use Commonwealth procurement and service delivery to improve outcomes and establish repeatable patterns for responsible AI deployment, measurement and assurance.

    Commonwealth procurement and service delivery can help improve outcomes for Australians while establishing repeatable patterns for responsible AI deployment, measurement and assurance.

  5. Have mandatory data-centre standards complement existing state and territory planning and environmental regimes.

    Standards covering energy, water, land use, infrastructure costs and skills can provide greater consistency while supporting community confidence and appropriate cost allocation. They should complement, rather than duplicate, existing state and territory planning and environmental regimes.

  6. Align energy/grid and land-use planning, environmental/development approvals, digital connectivity and infrastructure investment with credible developer pipelines in a nationally coordinated framework.

    The objective should be to create a nationally coordinated framework that provides both clear obligations and clear investment pathways. Governments should seek to align energy and grid planning, land-use planning, environmental and development approvals, digital connectivity and infrastructure investment with credible developer pipelines.

  7. Contribute to international standards, testing and assurance practice rather than simply adopting standards developed elsewhere.

    Australia should seek to contribute to the development of international practice in these areas rather than simply adopt standards developed elsewhere, and we welcome the Government’s plan to develop National AI Standards.

  8. Take a coordinated approach to planning, energy and investment processes to reduce regulatory complexity, improve investor confidence and deliver AI infrastructure at pace and scale.

    A coordinated approach to planning, energy and investment processes would reduce regulatory complexity, improve investor confidence and help ensure Australia can deliver projects at the pace and scale required.

  9. Recognise standards, testing and assurance capability as part of how Australia's sovereign AI capability is defined and developed.

    Recognise standards, testing and assurance capability as part of how Australia's sovereign AI capability is defined and developed.

  10. Consider how Australia could build, coordinate and prioritise sovereign AI expertise across research, commercialisation, standards, testing and assurance.

    Government to consider how Australia could build, coordinate and prioritise expertise in sovereign AI capability across research, commercialisation, standards, testing and assurance.

  11. Target capability development at exposed cohorts and sectors, encourage experimentation and continuous learning, and assess benefits through productivity, capability, job quality and economic inclusion.

    We suggest the central policy objective should be human-centred AI adoption. This means protecting access to entry-level pathways, targeting capability development at cohorts and sectors most exposed to change, encouraging organisational approaches that support experimentation and continuous learning, and measuring not just adoption on its own but who benefits. Success should ultimately be assessed by whether AI improves productivity, workforce capability, job quality and economic inclusion, particularly for cohorts at risk of being left behind by technological change.

  12. Protect entry-level pathways by encouraging education providers and employers to build judgement, communication, leadership and problem-solving capabilities earlier in workers' careers.

    Protect entry-level pathways by encouraging education providers and employers to build judgement, communication, leadership and problem-solving capabilities earlier in workers’ careers.

  13. Support flexible, industry-informed approaches to lifelong learning and reskilling that can respond to rapidly changing occupational requirements.

    Support flexible, industry-informed approaches to lifelong learning and reskilling that can respond to rapidly changing occupational requirements.

  14. Ensure public investment in AI skills develops technical proficiency alongside human capabilities including judgement, creativity, collaboration and ethical decision making.

    Ensure public investment in AI skills develops technical proficiency alongside complementary human capabilities, including judgement, creativity, collaboration and ethical decision making.

  15. Expand national workforce monitoring to capture participation, capability development, job quality and organisational transformation, not only rates of AI adoption.

    Expand national workforce monitoring to capture participation, capability development, job quality and organisational transformation — not simply rates of AI adoption.

  16. Incorporate AI capability into public-sector workforce planning, particularly where attracting, retaining and developing specialist skills is critical to adoption.

    Incorporate AI capability into public-sector workforce planning, particularly where attracting, retaining and developing specialist skills will be critical to successful adoption.

  17. Set agent identification, authorisation, monitoring and stopping rules, assign responsibility for actions beyond intended roles, and consider testing unfair bias where agents influence decisions affecting individuals.

    Clear rules are needed to ensure agents are identified, authorised, monitored and able to be stopped, with responsibility assigned when they act beyond their intended role. Agents could also be monitored and tested for unfair bias, including gender and cultural bias, particularly where they influence decisions affecting individuals.

  18. Provide consistent AI records and assurance evidence for boards and regulators while avoiding duplicate reporting.

    Organisations need a consistent way to record the AI they use and show that it is operating responsibly. A recognised assurance framework, aligned with international standards, would provide boards and regulators comparable evidence while avoiding duplicate reporting.

  19. Consider a proportionate material-service-provider framework for identifying and managing critical third-party AI dependencies, with appropriate provider transparency and assurance obligations.

    A proportionate framework could draw on prudential regulation’s concept of a material service provider, requiring organisations to identify and manage critical third-party AI dependencies while placing appropriate transparency and assurance obligations on providers.

  20. Preserve legal obligations in pre-deployment clearance, while clarifying expected controls and evidence.

    The Committee could also consider a pre-deployment regulatory clearance pathway through which an organisation could test a proposed AI deployment with the relevant regulators before launch and receive a documented view of the expected controls and evidence. This would not remove any legal obligation, but would clarify how obligations can be met.

  21. Close targeted gaps in existing frameworks first, prioritising supply chain accountability, agent identity and authorisation, and inventory and assurance obligations for high-risk uses.

    Close targeted gaps in existing frameworks first, prioritising supply chain accountability, agent identity and authorisation, and inventory and assurance obligations for high- risk uses.

  22. Establish a standing regulator coordination mechanism, which could sit within the Office of AI, so economy-wide AI expectations are set once and applied consistently by sector regulators.

    Establish a standing regulator coordination mechanism (which could sit within the new Office of AI) so economy-wide AI expectations are set once and applied consistently by sector regulators.

  23. Anchor any mandatory guardrails in a recognised assurance framework aligned to ISO/IEC 42001 and the NIST Risk Management Framework, and accept continuous automated control testing as evidence.

    Anchor any mandatory guardrails in a recognised assurance framework aligned to international standards ISO/IEC 42001 and NIST Risk Management Framework, and recognise continuous automated control testing as acceptable evidence.

  24. Establish coordinated pre-deployment regulatory clearance giving organisations documented regulatory views before launch and publishing de-identified determinations, drawing on international models.

    Establish a coordinated pre-deployment regulatory clearance pathway giving organisations a documented view from the relevant regulators before launch, with determinations published in de-identified form, drawing on various international models.

Read the original

Open in the interactive guide