Artificial Intelligence Accountability of Public Administration in Canada

With co-author Brandon Orct I prepared the country report on Canada for the “Artficial Intelligence Accountability” stream for this year’s General Congress of the International Academy for Comparative Law. Our report is available now on SSRN:

This is the country report on “Artificial Intelligence Accountability” prepared in respect of Canada for the International Academy of Comparative Law’s 2022 General Congress.

AI and algorithmic systems are being utilized by governments to make or assist in decision-making. Such systems are being used for administrative purposes such as determining eligibility for government benefits, conducting risk assessments, determining immigration statuses, and assisting in investigations and regulatory matters. These systems have also been deployed extensively in criminal justice and law enforcement practices in governments across the United States, the United Kingdom, and the European Union. Canada is no exception to this trend, albeit possibly relying on these systems at a lesser extent to other jurisdictions in the context of criminal justice. We discuss the Canadian trends in Part 1.

In Part 2, we turn to consider the accountability mechanisms which are currently in place. To date, the Canadian legal framework for regulating the use of such systems is underdeveloped. Soft law comprises the majority of algorithmic accountability in Canada at this current juncture. There is a considerable body of policies that can impact how administrative decision-making occurs. Such ethical guidelines, directives, playbooks or other best practices have significant potential to supplement hard law and other mandatory legal measures (if and when adopted) on algorithmic systems. Consequently, soft law has, and will likely continue to have, a strong role in shaping the law of algorithmic accountability. Canada’s extant body of privacy law imposes some constraints (on private and public sector actors alike) on the development and deployment of AI and algorithmic systems, by regulating the use of personal data. But there is an emerging consensus that Canadian privacy law needs to be updated to take account of the growth in AI and algorithmic systems. Relatedly, there has been lively debate about the possibility of creating specialized oversight bodies. And, at the margins, the judiciary may intervene, by way of judicial review, to limit the scope for the use of AI and algorithmic systems in public administration.

In the pages that follow, therefore, we survey, first, the use of AI and algorithmic systems in Canadian government. We then, second, critically analyze accountability mechanisms: soft law, privacy law, oversight bodies, and the judiciary.

You can download it here.

This content has been updated on November 2, 2022 at 20:02.