Submission 52
Accenture
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.
Summary
Read the original submission for the full text.
Accenture argues Australia should move from AI readiness to practical, responsible adoption, treating AI as national productivity and resilience infrastructure. It proposes a sovereign AI framework favouring domain-specific models, capability transfer in AI partnerships, an APS AI workforce strategy, whole-of-government adoption capability and responsible-AI controls, and cites its own products as case studies.
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.
Accenture argues AI should be treated as national productivity and resilience infrastructure rather than simply a new class of technology.
AI should be treated as national productivity and resilience infrastructure, not simply as a new class of technology.
Accenture argues sovereignty means a nation's ability to develop and deploy AI using its own infrastructure, data, models and talent while still drawing on global technology ecosystems where appropriate.
Accenture’s sovereign AI research defines sovereignty more broadly as a nation’s ability to develop and deploy AI using its own infrastructure, data, models and talent while still drawing on global technology ecosystems where appropriate.
Accenture argues narrow, domain-specific models will be more useful, governable and trusted than broad general-purpose models for many Australian use cases.
For many Australian use cases, narrow and domain-specific AI models will be more useful, governable and trusted than broad general-purpose models.
Accenture argues major AI partnerships should leave a capability dividend in Australia, so local agencies, businesses and workers can understand and maintain the models.
major AI partnerships should leave a capability dividend in Australia. Sovereignty in nationally important AI systems depends not only on secure data arrangements, but on whether Australian agencies, businesses and workers can understand how models are trained, tested, governed and monitored, and maintain or adapt them over time.
Accenture argues public servants need skills to commission, govern, challenge and explain AI systems, beyond using them safely.
Public servants must be able not only to use AI safely, but to commission, govern, challenge and explain AI systems used in public administration and support of citizen services.
Accenture argues responsible AI principles must shape delivery decisions, with assurance from inception to retirement because risks often become visible after deployment.
Responsible AI principles will only protect the public if they shape delivery decisions. Accenture’s responsible AI work highlights that risks often become most visible after AI deployment, so assurance must operate from inception to retirement rather than as a one-off pre-launch check.
Recommendations
Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.
The Committee should judge AI settings by their contribution to removing economy-wide bottlenecks, sovereign capability and resilience, competitiveness and living standards.
The Committee should therefore judge Australia’s AI settings by whether they help remove economy-wide bottlenecks, strengthen the nation’s sovereign capability and resilience, lift business and economic competitiveness, and improve living standards.
Define the data, assurance, workforce and partnership foundations for priority sectors and use cases, and measure real productivity and resilience gains.
This requires a practical national AI adoption agenda that identifies priority sectors and use cases, defines the data, assurance, workforce and partnership foundations needed for each, and measures whether AI is delivering real productivity and resilience gains.
Establish a national AI productivity and resilience agenda with a small number of nationally significant adoption priorities and a reusable evidence base of tested use cases, assurance requirements, lessons and public value measures.
The Australian Government should establish a national AI productivity and resilience agenda, identifying a small number of nationally significant AI adoption priorities and building a reusable evidence base of tested use cases, assurance requirements, implementation lessons and public value measures.
Develop a sovereign AI framework identifying priority nationally important sectors and supporting domain-specific models on trusted Australian data where this improves accuracy, safety, contestability and alignment with Australian law, policy and public expectations.
The Australian Government should develop a practical sovereign AI framework that identifies priority nationally important sectors and supports narrow, domain-specific models trained or tuned on trusted Australian data where this improves accuracy, safety, contestability and alignment with Australian law, policy and public expectations.
Adopt risk- and values-based sovereignty: use global-scale tools where appropriate, requiring higher sovereign control for sensitive personal data, critical infrastructure, national security, regulatory decisions, social entitlements or areas where public trust and accountability are paramount.
Instead, the Government should adopt a risk-based and values-based approach, using global-scale tools where appropriate, while also requiring higher sovereign control where AI interacts with sensitive personal data, critical infrastructure, national security, regulatory decisions, social entitlements or other areas where public trust and accountability are paramount.
Identify the key Australian datasets, laws, policies and regulatory guidance needed for trusted AI, and create safe access arrangements with clear rules for testing, auditing, correcting errors and monitoring performance.
The Australian Government should identify the key Australian datasets, laws, policies and regulatory guidance needed for trusted AI, and create safe access arrangements with clear rules for testing, auditing, correcting errors and monitoring performance.
Build local skills and research collaboration through AI partnerships/procurements; transfer knowledge to public servants and regulated industries, enable public-interest compute access, strengthen model/data governance and create Australian supply-chain opportunities.
Partnerships and procurements should therefore build local skills, support Australian research collaboration, transfer knowledge to public servants and regulated industries, provide appropriate access to compute for public interest use cases, strengthen model and data governance, and create opportunities for Australian firms to participate in AI supply chains.
Require major AI partnerships and procurements to demonstrate capability transfer, Australian skills, secure data, model assurance, research collaboration and transparent public value.
The Australian Government should require major AI partnerships and procurements to demonstrate capability transfer, Australian skills development, secure data arrangements, model assurance, research collaboration and transparent public value outcomes.
Make AI comprehension and application baseline national workforce capabilities; support responsible work redesign through worker/operator co-design, clear human oversight, practical training and measures of productivity, job quality and staff experience.
The Australian Government should make AI comprehension and application a baseline national workforce capability and support employers to redesign work responsibly, including through worker and operator co-design, clear human oversight, practical training and measures of productivity, job quality and staff experience.
Connect the national AI adoption approach with job redesign, consultation, skills mobility and transparent measures of productivity, quality, risk reduction and staff experience.
A credible national approach should therefore connect AI adoption with job redesign, consultation, skills mobility and transparent measures of productivity, quality, risk reduction and staff experience.
Develop an APS AI Workforce Strategy covering baseline comprehension and application, specialist technical capability, AI-enabled service design, product management, model risk, procurement and responsible AI assurance.
The Australian Government should develop an APS AI Workforce Strategy covering baseline comprehension and application, specialist technical capability, AI-enabled service design, product management, model risk, procurement and responsible AI assurance.
Distinguish all-staff literacy, manager and executive capability, specialist data engineering, assurance, cyber, procurement, product management, service design, legal risk and evaluation skills, and senior leadership capability balancing innovation, efficiency, fairness, risk and public trust.
A mature strategy should separate all-staff AI literacy from manager and executive capability, specialist skills in data engineering, model assurance, cyber security, procurement, AI product management, service design, legal risk and evaluation, and senior leadership capability to balance innovation, efficiency, fairness, risk and public trust.
Move towards intelligent government using a human-centred, transparent, contestable approach aligned with public value.
Australia should take the same direction in a distinctly Australian way, using a human-centred, transparent, contestable and public value-aligned approach.
Move agencies to mission-led AI delivery with shared assurance tools, model registers, implementation patterns, procurement guidance, training and benefits measurement; allow agencies to adapt common foundations without centralising every decision.
This does not require centralising every decision. It requires building common foundations once, then allowing agencies to adapt them. Recommendation 7: The Australian Government should move from agency-by-agency pilots to mission-led AI delivery, supported by a whole-of-government AI adoption capability that provides common assurance tools, model registers, reusable implementation patterns, procurement guidance, training resources and benefits measurement support.
Agencies should determine appropriate AI uses, human oversight, output testing and challenge, accountability, when manual control is required and how systems will be monitored.
Agencies need to know which problems are appropriate for AI, what human oversight is required, how outputs will be tested and challenged, who is accountable, when manual control is required and how systems will be monitored.
Give citizens clear explanations of why AI is used, its role, how decisions can be reviewed and who remains accountable.
Citizens also need clear explanations of why AI is being used, what role it plays, how a decision can be reviewed and who remains accountable.
Require responsible AI controls across procurement, service design, data governance, cyber assurance, testing, training, monitoring/executive decisions for significant government deployments, with accountability, human review, audit trails and plain-language explanations for high-impact or public-facing uses.
The Australian Government should require responsible AI controls to be embedded across procurement, service design, data governance, cyber assurance, testing, workforce training, monitoring and executive decision-making for significant government AI deployments, with clear accountability, human review rights, audit trails and plain-language public explanations for high-impact or public-facing uses.