7 Domains of AI Readiness

AI capabilities, in particular generative AI tools, are being rapidly commodified.

Morgan Stanley argue over 40% of the labour force could be affected by AI within 3 years. Government agencies are already reflecting on impacts for the people and industries they service, but what should they be acting on now to prepare for changes to their own operations?

Public sector agencies build operating models to deliver on their legislative and policy mandate through fair, timely, defensible & intended outcomes. Each element of this operating context should be considered when deciding how the agency will approach AI adoption.

1. Legislation and Mandate

Legislation provides a top-down guide for what and how AI technologies can be implemented. Agencies will need to consider their enabling legislation, privacy, anti-discrimination legislation, the PGPA Act and other relevant legislation. 

Australia's legal environment is expected to continuously evolve to accommodate the use of AI. As a starting point, legislative frameworks in the EU and Canada offer valuable guidance.

2. Strategy, Policy and Ethics

Agencies should provide top-down scope and constraints within strategic documents for use and exploration of AI capabilities, linked directly to intended outcomes, refreshed risk frameworks, and integration of AI policy and ethics guidance.  Implementation and governance approaches should be guided by risk.

Important sources include the DTA’s Interim Guidance for agencies on government use of generative artificial intelligence platforms, DISR’s Safe and Responsible AI discussion paper and Australia’s AI Ethics Principles. 

3. Governance and Controls

Agencies face big questions. Answers should guide implementation of whole-of-lifecycle AI governance and risk management practices. 

1. What do these technological advances mean for the services and functions we deliver to our stakeholders?

2. What can we leverage?

3. What can we leave?

4. How do we utilise these opportunities to be more efficient and effective in pursuing our vision?

5. What is our risk appetite?

6. What further opportunities and risks are coming down the pipe?

7. What does this mean for our operating model now, in 6 months, in two years?

8. How do we govern uptake, execution, risk and opportunity?

4. Service Model

An agency’s service model may change over time with the adoption of AI capabilities, evolutions in agency strategy, and in response to stakeholder demand.

Changes should be made iteratively based on tested and assured capabilities, whole-of-lifecycle service delivery governance and well-defined risk management practices.

5. Technology, Information and Data

AI capabilities are already commoditised within Microsoft stack products, and increasingly, domain-specific tools feature AI capabilities, impacting many spheres of government work.

Agencies will need to assess the capabilities most likely to add value in their environment both now and into the future. This will include an assessment of enterprise capabilities and data holdings. 

This bottom-up exploration should inform strategy design.

6. Processes and Procedures

AI capabilities are likely to drive process efficiencies – but these should be explored carefully, and within a risk and governance regime that priorities human benefit, fairness, privacy and security, reliability, transparency and explainability, contestability and accountability.

Agencies are likely to undertake an iterative, bottom-up discovery journey concerning AI-driven process improvements. This journey should be guided by a clear framework that weighs benefits and risks over the short, medium, and long terms

7. People, Culture and Training

AI capabilities will change the way teams work – opening pathways for more streamlined regulatory, service delivery, policy and program delivery functions.

This is likely to change the way teams work together and interact with stakeholders. The agency should carefully consider people, change and teaming shifts with the adoption of new technologies and feed assessments systematically into AI adoption strategies and plans.