Agreement Reached on National AI Framework by Data and Digital Ministers

Australia's federal government has created a taskforce dedicated to ensuring the safe and responsible use of artificial intelligence (AI) within the public service, aiming to maximize the benefits of AI technology for the broader community. This initiative is supported by the AI in Government Taskforce, along with interim guidance introduced last year to facilitate the transformation of service delivery to Australians. Complementing these efforts, an AI expert group has been established, implementing an action plan that has been in effect since 2021. Various federal and state agencies across Australia are actively exploring the implementation of AI technology.

One notable example is the Victorian Office of Public Prosecutions (OPP), the state’s largest criminal legal practice, which is implementing a case management system on the Appian AI Process Platform to enhance its operations. This solution leverages Appian’s data fabric and automation technology to accelerate the resolution of criminal matters, driving efficiency and improving the resolution process experience for employees and crime victims. By streamlining operations, reducing administrative burdens, and automating processes, the OPP is well-positioned to enhance productivity and optimize service delivery.

Case management is a particularly impactful area for government agencies to enhance with AI. Government applications in this realm are diverse, ranging from building consents and permitting to collecting evidence for prosecuting criminal cases, as seen with the Victorian OPP. Additionally, other areas where Australian government agencies could benefit from AI include food and drug safety, where multiple steps with various documents and stakeholders, such as officials that approve applications and permits, are involved.

Advanced large language model (LLM) AI systems can significantly aid these processes by understanding, analyzing, interpreting, and generating human language. These systems use deep learning techniques to predict, produce, and output coherent text in response to user prompts. An LLM trained on government agency data can perform tasks like text extraction, translation, summarization, and conversational responses. These capabilities provide intuitive interactions with data or documents through natural language queries, allowing users to ask questions through chat and receive answers based on the data.

While AI can enhance efficiencies, its use in government workflows raises significant privacy and security concerns. Ensuring that data remains confidential and is handled in compliance with regulations and policies is crucial to maintaining public trust and national security. Using public AI engines that incorporate all available data for training runs the risk of exposing sensitive government information. This highlights the importance of using private AI models to ensure government data remains confidential and protected.

Appian has addressed these concerns in its Case Management Studio for Public Sector solution, which uses a private AI model to keep data proprietary. This model is trained directly on the organization’s data, allowing the organization to retain full control and ownership over its sensitive information and safeguarding it from unauthorized access or exposure. Through a strategic collaboration with AWS, leveraging Amazon Bedrock and Amazon SageMaker, Appian is able to host LLMs within customer compliance boundaries on the Appian Platform, ensuring data remains secure. All data hosted on the Appian cloud meets the data residency requirements of Australian government agencies, with all data used during development, production, and disaster recovery located in Australia.

To maximize AI's potential in public service, it is imperative for government bodies to educate their staff on new AI tools as they are introduced. Leading this effort is Appian, which is at the forefront of integrating AI into Australian public service operations.

In addition to these specific implementations, the Data and Digital Ministers have agreed on a set of nationally consistent approaches to ensure the safe and ethical use of AI in government projects and programs. Commonwealth, state, and territory governments have agreed to and released the National Framework for Assurance of AI in Government, following a meeting in Darwin, according to a joint statement released by the Digital and Data Ministers Meeting (DDMM) group.

Senator Katy Gallagher, DDMM chair and Minister for the Australian Public Service, stated that all levels of government agreed that the rights, wellbeing, and interests of people should be prioritized whenever a jurisdiction considers using AI in policy and service delivery. The set of guidelines, best practices, and standards are based on the federal government’s eight AI ethics principles, which are promoted across the public and private sectors. These principles include human, societal, and environmental wellbeing; human-centered values; fairness; privacy protection and security; reliability and safety; transparency and explainability; contestability; and accountability.

NSW Minister for Customer Service and Digital Government, Jihad Dib, emphasized that the framework provides flexibility for a jurisdiction’s unique needs while defining consistent expectations for oversight of AI and people's experience of government. NSW was the first state to mandate internal assessments for public sector AI projects and external reviews for projects that exceed defined risk thresholds in 2022. Following this, Western Australia became the second jurisdiction to establish AI-specific risk assessments for public sector projects in March 2023.

The national framework encourages governments to consider similar auditing processes and oversight mechanisms for high-risk settings, such as external or internal review bodies, advisory bodies, or AI risk committees. These bodies provide expert advice and recommendations to ensure AI use is responsible and ethical. Additionally, governments are encouraged to assess AI use cases through impact assessments, evaluating the likely impacts on people, communities, and the environment to determine if benefits outweigh risks and to manage these impacts appropriately.

The establishment of a national framework for AI governance in Australia's public sector marks a significant step towards the safe and ethical use of AI technologies. By adhering to the federal government's AI ethics principles and implementing rigorous oversight mechanisms, government agencies can harness the transformative potential of AI while safeguarding the rights and wellbeing of individuals. The collaborative efforts of federal and state governments, coupled with robust privacy and security measures, will ensure that AI serves as a force for good, enhancing service delivery and improving outcomes for all Australians.