The government has been using AI… a lot

Much like a lazy student tired of writing essays, the federal government loves using AI. 

Driving the news: The feds have used AI in nearly 300 projects and initiatives, per new research from a professor at Western University. It took two years to compile all the use cases, scraping info from databases, news sources, and access-to-information requests.

  • The Canadian Food Inspection Agency uses AI to track invasive plants and insects, while Fisheries and Oceans Canada uses it to detect marine mammals in images.

  • Other use cases are more serious. The RCMP and border agency both use facial recognition technology, a contentious tool some advocates want banned outright

Why it matters: The report raises concerns about government transparency regarding new and powerful AI tools and their use in highly sensitive areas. It’s especially troubling given that proposed AI protection legislation would mostly not apply to the government’s use of AI. 

  • The Artificial Intelligence and Data Act promises consumer protection from businesses using “high-impact” AI systems, but some critics argue it lacks similar language applying to governmental use.

  • Examples of so-called high-impact systems include biometric identification and employment screening, a.k.a. two things the feds have already used AI for.

Bottom line: The feds are trying to walk a fine line between letting AI innovation flourish (hey there, $2.5 billion pledge) and reining in harms. So far, these harms have been focused on what companies might do with AI. It could be time to look inward.—QH