AI system lifecycle

The practices described in the standard use a reference AI lifecycle model to ensure holistic coverage of an AI system from inception to retirement, as shown in the 'AI lifecycle diagram' below.

The statements and criteria outlined in this standard are structured according to the relevant lifecycle stages and are intended to be implemented through an iterative process.

The AI system lifecycle is a structured process that occurs in stages, ensuring the holistic coverage of the AI system from discovery to retirement.

The AI lifecycle stages include:

  1. Discover: design, data, train and evaluate.
  2. Operate: integrate, deploy and monitor.
  3. Retire: decommission.

This lifecycle model is based on the  Voluntary AI Safety Standard.

AI lifecycle diagram

AI system development is generally an iterative approach. At any point of the lifecycle, issues, risks, or opportunities may be discovered for improvement that could prompt changes to system requirements, design, data, model, or test cases. After deployment, feedback and issues could prompt changes to the requirements.

Each agency may have existing architecture and processes relating to the adoption and implementation of AI systems. The standard complements existing architecture and processes.

The Policy for the responsible use of AI in government encourages continuous improvement to enable AI capability uplift.
 

Applying the lifecycle and standard requirements

AI system development is generally an iterative approach. At any point of the lifecycle, issues, risks, or opportunities may be discovered for improvement that could prompt changes to system requirements, design, data, model, or test cases. After deployment, feedback and issues could prompt changes to the requirements.

Each agency may have existing architecture and processes relating to the adoption and implementation of AI systems. The standard complements existing architecture and processes.

The Policy for the responsible use of AI in government encourages continuous improvement to enable AI capability uplift.

Connect with the digital community

Share, build or learn digital experience and skills with training and events, and collaborate with peers across government.