`Addressing AI governance challenges in Canada, experts emphasize the need for collaborative regulation to balance innovation and public trust. (IT World Canada)
A recent conference on artificial intelligence (AI) in the public sector emphasized the need for comprehensive regulation of AI across all levels of government in Canada, echoing concerns about the unchecked use of AI in the private sector. Stephen Toope, CEO of the Canadian Institute for Advanced Research (CIFAR), cautioned against isolated national or provincial regulations, suggesting the necessity of forming "regulatory coalitions" with other jurisdictions, particularly with the European Union, to avoid a race to the bottom in regulatory standards.
Toope acknowledged the complexity of achieving such coordination but emphasized its importance in maintaining public trust and fostering innovation. He criticized the feasibility of a global compact around AI proposed by some CEOs, deeming it a cynical exercise unlikely to materialize.
The conference, organized by Ontario’s Information and Privacy Commissioner during Data Privacy Week, opened with calls for a binding AI framework in the province. Patricia Kosseim, the Commissioner, reiterated the need for rules governing AI use in the public sector. Melissa Kittmer, assistant deputy minister in Ontario’s Ministry of Public and Business Service Delivery, shared that the government has been developing a Trustworthy AI Framework since 2021, focusing on transparency, accountability, and responsible AI use.
While Ontario is already employing AI in various sectors, including data extraction, chatbots, virtual assistants, and predictive soil mapping, Kittmer did not specify a release date for the framework.
The discussion touched on ongoing legislative initiatives related to AI across the country, including the proposed Artificial Intelligence and Data Act (AIDA) being debated in Parliament. However, AIDA only covers federally regulated businesses and firms in provinces without their own AI legislation. To address the challenges posed by AI, Ottawa issued a directive in 2019 for federal civil service AI use, with a guide for the use of generative AI released in the following year.
Comparatively, the European Union is making progress toward a comprehensive AI Act that covers both private and public sectors, with hopes for its passage before the summer elections.
Kosseim highlighted the potential benefits of AI in government, such as drafting plain language summaries, improving access to government services, enhancing healthcare diagnostics, and aiding emergency services. However, she cautioned against the inherent risks, citing instances of biased algorithms perpetuating discrimination.
Toope shared a story of a Canadian Black computer scientist encountering facial recognition bias, emphasizing the need for diverse representation in AI system development. Chris Parsons from the Ontario privacy commissioner's office noted that while big companies are aware of AI challenges and implement safety checks, the underlying data can still introduce bias. He urged organizations to turn to guidance from privacy commissioners while waiting for comprehensive laws.
The discussion extended beyond personal privacy, with concerns raised about AI's impact on non-personal information and its potential to categorize communities, affecting the delivery of services. Calls were made for a broader focus on the societal impact of technology, advocating for "group rights" to empower individuals against powerful entities deploying AI.
