Decoding Canada’s Directive on Automated Decision-Making: A blueprint for AI ‘guardrails’?
Author ADM+S Centre
Date 31 May 2024
The pace at which advances in generative AI are being made accessible by companies, without perceived oversight, has sharpened the focus of governments worldwide on ensuring there are sufficient ‘guardrails’ for the development and deployment of AI.
It’s a good time to assess potential approaches to AI regulation in Australia.
The Safe and Responsible AI in Australia discussion paper released by the Federal Government last year proposed a risk-based approach, focused on setting up additional guardrails to reduce the likelihood of harms occurring in high-risk settings in the development and deployment of AI.
The discussion paper gave extensive consideration to Canada’s Directive on Automated Decision-Making (ADM) which focuses on processes that encourage fairness, accountability and transparency in government decision-making, rather than prohibiting particular use cases or outcomes.
In a recent article Decoding Canada’s Directive on Automated Decision-Making: A blueprint for AI ‘guardrails’? Research Fellow Dr Henry Fraser from the ARC Centre of Excellence for Automated Decision-Making and Society and QUT Law students Jacqueline McIlroy and Sara Luck discuss the Canadian Directive on ADM and what it reveals about the strengths and limitations of a ‘guardrails’ approach to AI regulation.
“Deciding which risks from AI are acceptable, and which are not, is incredibly challenging,” writes Fraser, McIlroy and Luck.
“The Directive offers a solid starting point for AI regulation, engaging deep policy questions about rights, safety, efficiency, public interests, and social justice.”
To illustrate the practical application of these guardrails, the authors imagine how the ADM Directive’s requirements might have impacted the notorious Robodebt system.
As Australia moves forward in shaping its AI regulatory framework, the lessons from Canada’s ADM Directive provide a valuable blueprint.
