I have continued to write about different aspects of AI and legal judgement, and in the latest issue of Constitutional Forum, I have an article titled “Is There a Constitutional Right to a Human Decision in Canada?”
The abstract reads:
In the mid-2010s, government interest in and uptake of automated decision-making prompted debate over law and policy on point. At issue were forms of AI that did not give reasons and relied on algorithms of varying degrees of opacity. Canada’s Treasury Board Directive on Automated Decision-Making and Guide on the Use of Generative Artificial Intelligence follow the EU’s GDPR and AI Act in recognizing rights to a human decision and to an explanation of automated decision-making, thereby limiting the use of AI in high-impact cases. But recent advances in AI challenge the assumptions underlying this approach. Language models can generate reasons for a decision comparable in quality to those of a human — prompting a need to look beyond existing law and policy toward constitutional limits on use of this technology. This article outlines the contours of Charter rights to a human decision and an explanation, which, for the foreseeable future, preclude significant reliance on language models in high-impact administrative cases regardless of their capabilities.
The paper can be found here . ■