What is Machine Learning?
Machine learning (ML) is a field within artificial intelligence (AI) and computer science dedicated to using data and algorithms to help AI mimic human learning, thereby enhancing its precision over time.
While there are various definitions of what constitutes artificial intelligence, the DTA and AGA use the OECD definition of an Artificial Intelligence (AI) system.
An Artificial Intelligence (AI) system is a machine-based system that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments. Different AI systems vary in their levels of autonomy and adaptiveness after deployment.
Agencies may refer to explanatory material on the OECD website.
Given the rapidly changing nature of AI, agencies should keep up to date on changes to this definition. The definition may be reviewed as the broader, whole-of-economy regulatory environment matures to ensure an aligned approach.
There may be instances, such as considering whether to apply AI assurance processes, where agencies may wish to provide practical guidance to staff to identify AI use.
Purpose
A purpose statement specific to Machine Learning will be finalised through the iterative development of the Australian Government Architecture. The below is considered applicable across the Domain of Artificial Intelligence.
Artificial Intelligence makes it possible for machines to learn from experience, adjust to new inputs and perform human-like tasks. This has the potential to increase the efficiency and accuracy of entity operations, allow improved exploration of data and derivation of insights, and improve service delivery for people and business.
The capability of AI is realised through:
- deployment of AI technologies to support Commonwealth entities’ unique needs
- alignment with Australia's AI Ethics Principles
- being an exemplar in the safe and responsible use of AI, requiring a lawful, ethical approach that places the rights, wellbeing and interests of people first
- adopting AI assurance practices that align best practice including those within the National framework for the assurance of artificial intelligence in government and developing Australian government and whole of economy safe and responsible AI initiatives
- continual review and improvement of AI use and practice, in recognition of its state as an emerging technology area.
Objective
Objectives specific to Machine Learning will be finalised through the iterative development of the Australian Government Architecture. The below are considered applicable across the Domain of Artificial Intelligence.
The objectives of this Australian Government Architecture (AGA) content are to:
- ensure that entities engage with AI confidently, safely and responsibly, and realise its benefits
- strengthen public trust in government’s use of AI
- ensure strategic alignment of the adoption of AI to the Australian Government’s data and digital goals
- meet compliance obligations with legislation and regulation, government policies and standards, and relevant national or international agreements relating to AI
Whole of Government Applicability
The Australian Government has published its interim response to the safe and responsible AI consultation held in 2023. The response considers the below principles that should be paramount to any entities' own considerations:
- Risk-based approach
- Balanced and proportionate
- Collaborative and transparent
- A trusted international partner
- Community first
The use of AI by Commonwealth entities has the potential to contribute to the seamless delivery of government services across different systems and processes. To do so, entities should consider:
- reuse of commercial (including whole-of-government) arrangements
- replication and redeployment of proven AI solutions
- mobility of APS employees to support knowledge sharing
- reuse of lessons learned from prior implementations.
The Data and Digital Government Strategy and Implementation Plan set directions to the APS for AI through:
- Delivering for all people and business: To maximise value from data
- Simple and seamless services: To deploy scalable and secure architecture
- Government for the future: To adopt emerging technologies
- Trusted and secure: To connect data, digital, and cyber security, and build and maintain trust
Policy Elements
-
Designate accountability to accountable official(s)
Designate accountability for implementing the policy to accountable official(s) within 90 days of the policy taking effect.
Requirements for designating accountable official(s) are set out in the Standard for accountable officials.
-
Publish an AI transparency statement and keep it updated
Make publicly available a statement outlining their approach to AI adoption and use within 90 days of the policy taking effect.
Review and update AI transparency statement annually or sooner, should the agency make significant changes to their approach to AI.