A recent study found that Canada’s federal government has utilized artificial intelligence in nearly 300 projects and initiatives in various departments, including predicting tax case outcomes, sorting temporary visa applications, and promoting diversity in hiring. The data was compiled by Joanna Redden, an associate professor at Western University, who used news reports, documents tabled in Parliament, and access-to-information requests to piece together the database. Redden argued that there needs to be more public debate and information about the use of AI systems in government, highlighting an issue with the proposed Artificial Intelligence and Data Act, which may not apply to most government uses of AI.
Some of the federal agencies that have used AI include the Department of National Defense, which implemented AI to reduce bias in hiring decisions and the Immigration, Refugees and Citizenship Canada, which used AI tools to sort visa applications. While some experiments, like a project analyzing social media for signs of suicide by the Public Health Agency of Canada, were discontinued due to factors like cost and methodology, others have become permanent initiatives. Health Canada continues to use a social listening tool with an AI component to monitor news for incidents related to consumer products.
The Royal Canadian Navy has experimented with voice-activated technology on warships, similar to Apple’s Siri or Amazon’s Alexa, with considerations for information security concerns before full implementation. The Canada Revenue Agency utilizes AI to predict court outcomes in tax cases, while the Canadian Institutes of Health Research uses labor relations decisions software to simulate different scenarios and their outcomes. Other applications of AI include flagging anomalies in estate filings at the Office of the Superintendent of Bankruptcy and identifying child sexual assault material by the RCMP.
Facial recognition technology has been implemented by federal agencies like the RCMP and the Canada Border Services Agency, with the latter using it on a voluntary basis to authenticate the identities of incoming travelers at airports. Joanna Redden has raised concerns about facial recognition technology and the need for better tracking of government AI use. While the government has a directive on automated decision-making that requires algorithmic impact assessments for AI systems with significant impacts, only 18 assessments have been published in a public register, which is smaller than Redden’s total findings.
Various departments like the Canadian Food Inspection Agency and Fisheries and Oceans Canada have utilized AI to track invasive species and detect marine mammals from aerial imagery. However, compiling this data was challenging for Redden due to limited information available about the AI systems, including when they were introduced, issues with the systems, or if they are still in place. Some pilot projects documented in the study, such as an algorithm for postal X-rays to detect guns by the Canada Border Services Agency, have not been updated, demonstrating the lack of transparency and accountability in government AI use. Overall, the study highlights the widespread use of AI in Canadian government bodies and the need for more information and oversight.
