The AI CoLab
A cross-sector partnership to accelerate responsible and ethical AI in the public sector
Why the CoLab Exists
- Accelerating AI in government safely is like planning a city - we need different experts working together.
- We need advocates of change to define problems and set goals, industry tech wizards to build solutions, and academics and civil society groups to flag risks and uncover opportunities.
- Open innovation is essential as community voices will help to ensure AI solutions that impact diverse groups are effective, inclusive, and fair.
- Without a neutral space to collaborate, government agencies, different states, and non-government experts are all reinventing the wheel.
- This leads to wasted effort, missed chances to learn from each other, and a patchwork of different AI approaches that is difficult to reconcile later.
- We need a place where everyone can share ideas and tackle common challenges together from the start, through hands-on co-design and iterative testing.
- Those responsible for risk inside government agencies rightly demand evidence that the risks from AI systems are well managed before they are deployed widely.
- However, without a controlled, collaborative testing environment, this creates a catch-22 that stifles government's ability to identify and manage these risks in the first place.
- It's like learning to swim – you can't improve without getting in the water, so we need a safe pool to practice without fear.
- If the public sector doesn't work openly or experiment, the gap will grow between what AI can do and what government can use, understand, and control.
- It's like having a smartphone but only using it to make calls.
- If this continues, the public sector risks becoming outdated in how we make policies, serve the public, and ensure that AI tools impacting public services are safe and effective.
- Citizens with public sectors that act slowly on AI aren't safer. They will be more vulnerable as agencies become reliant on external contractors who keep pace with AI, but lack public accountability.
- Citizens expect their governments to play a driving role in shaping public purpose AI tools, ensuring they're not just safe and compliant, but fully and fundamentally aligned with the public interest.
- A public sector taking a passive approach to AI is like riding in the back of a taxi in an unfamiliar city – you're moving, but you're not deciding the route, the speed, or the destination.
Vision and Mission
Vision
To be the catalyst for responsible public sector innovation, driving more integrated, adaptive, and productive governments that better support the citizens they serve.
Mission
The AI CoLab's mission is to accelerate the adoption of responsible AI across the Australian Public Service by:
- Fostering cross-sector collaboration between government, academia, and industry
- Facilitating safe and responsible experimentation with AI technologies
- Developing and piloting practical, scalable AI solutions for complex policy challenges
- Sharing common design patterns, best practices, and lessons learned
Priority Streams: The ANVIL Model
The AI CoLab focuses on five priority streams, encapsulated in the ANVIL model, to drive innovation and responsible AI adoption:
- Anticipate
- Navigate
- Ventilate
- Integrate
- Liberate
Core Elements
Cross-Sector Partnership
- A cross-sector initiative with partners, not a government body with contractors
- Maintains close ties with government while retaining independence and agility
- Involving communities and civil society groups ensures inclusive, ethical design and citizen-centric goals
Shared Digital Environment
- Secure, collaborative platform for AI experimentation and development
- Controlled space for safe experimentation, driving safe, responsible, effective AI tools through iterative design
- Enables cross-agency and cross-sector collaboration in a controlled setting
Neutral Physical Space
- Headquartered on neutral ground in the ACT
- Satellite nodes in major cities to facilitate nationwide collaboration
- Designed as a neutral ground for government, industry, academia, and community stakeholders