Can AI make the government more efficient? Canada plans to find out

On a rainy morning at the University of Waterloo in Ontario, Treasury Board president Ginette Petitpas Taylor announced a new plan for incorporating artificial intelligence into the workings of the federal public service.

The March 4 announcement—which was largely drowned out by the kickoff of the U.S.-Canada tariff war that day—quietly signaled a transformation ahead for the federal public service that could revolutionize the way Canadians interact with their government, when they claim federal benefits or apply for a passport.

The Artificial Intelligence Strategy for the federal public service proposes to “unlock capabilities beyond human limits” by developing and harnessing AI tools to enhance public sector workers’ productivity and capabilities. The strategy identifies specific areas AI could be used in government including immigration, agriculture, and services like passport renewals, with a focus on developing tools like natural language processing and translation.

“We’re laying the groundwork to leverage the potential of this technology to deliver the best possible services to Canadians,” Petitpas Taylor said at the announcement.

“I think it’s a step in the right direction,” said Graham Dobbs, an economist at Toronto Metropolitan University’s The Dais think tank, who researches AI technology and its potential government-related uses. According to research that Dobbs has conducted but not yet released, many public-sector jobs are especially well-suited for AI tools, which can take over routine tasks and make workers more productive.

Governments around the world are integrating AI software into their operations and service delivery in a bid to improve efficiency and productivity, save money, improve outcomes, and make dealing with bureaucracies feel—well, less bureaucratic. The city of Memphis, Tennessee, uses AI-powered cameras mounted on its vehicles to identify potholes that need fixing. In Estonia, public-sector AI tools help manage traffic, coordinate data sharing to improve health-care delivery, and chat with citizens to streamline their interactions with government departments. Meanwhile, the United Kingdom’s ambitious AI strategy aims to save £45 billion ($83 billion) a year by automating processes currently performed by humans (despite several “false starts” so far).

Here in Canada, government employees have been using AI tools for years. The Treasury Board cited several ways in response to questions from The Hub: Immigration, Refugees and Citizenship Canada uses automated decision-making to speed up passport application processing; Agriculture and Agri-Food Canada offers a generative AI chatbot called “AgPal” that provides advice about support programs and services for those in the agricultural sector; Natural Resources Canada has a repository of around a dozen projects that use AI technologies, including algorithms that help predict where valuable mineral deposits could be located, and an AI tool that integrates groundwater data from multiple sources.

In addition, many departments are using large language model tools (or LLMs) to support their work, primarily for drafting content and summarizing meetings.

In an interview with The Hub, Stephen Burt, the chief data officer for the Government of Canada noted that government officials feeding documents into AI tools is helping them learn “good Canadian English and good Canadian French.”

Under the new strategy, federal departments will be required to identify and disclose the ways they’re already using AI, as well as identifying additional uses in development, and potential future uses.

“The strategy is intended to provide guidance on using AI tools to improve services for Canadians, streamline administrative processes, reduce repetitive tasks, and allow public servants to focus on higher-value activities,” wrote the Treasury Board in a statement responding to questions from The Hub.

While last year’s federal budget included plans to reduce the public service by 5,000 employees over four years, the Treasury Board insisted the AI plan is not intended to put federal employees out of work. “The objective…is to enhance the work of public servants, not replace them.”

Dobbs noted that AI technology generally isn’t mature enough to replace humans yet anyway.

Senior bureaucrats in charge of writing and implementing the plan said the government wants to move forward more quickly with AI adoption, in an attempt to keep up with the pace seen in other countries from Estonia to the U.K. to Singapore. But they add that Canada is not lagging far behind its competitors.

“I think on the public sector side, Canada is a leader in AI among governments. We’ve been very quick to move in this space,” said Burt.

But will the change prove too quick for the public? Surveys show many Canadians remain skeptical of AI—more so than residents of other countries, despite AI having largely been a Canadian invention. Critics of AI warn that the technologies that fall under the term are largely unproven and often error-prone, can be used for mass surveillance, and can amplify the potential for biases in automated decision-making.

Others decry a lack of transparency in how AI is already being used by governments across Canada. The Treasury Board will maintain safeguards to address concerns like these, Burt said. It plans to create a public-facing dashboard to increase transparency around where AI is being used, how personal data is being handled, and the progress of the strategy’s implementation.

One critic of government transparency around AI use is Canadian researcher Joanna Redden, co-director of the U.K.-based Data Justice Lab, which has released a data set tracking around 300 uses of AI within the Canadian federal government. Much of it is information that hasn’t been readily available anywhere else, said Redden, who is also co-director of the Starling Centre at Western University in London, Ont.

Redden praised a few aspects of the federal strategy, including the mentions of a public register disclosing government use of AI—which would create more transparency—and an ethics review board. However, she said, “It is not clear what measures and resources will be introduced to ensure that affected individuals and groups can challenge AI applications that affect them.

“One of the things that comes out of the research that we’ve been doing is that there is too often a focus on the hype around AI. There’s often a lot of discussion about the opportunities that may come with AI, and there’s some mention of risk, but risk tends to be treated as if it’s a potential future prospect” rather than a present risk, she said.

In the midst of a trade war with the U.S., some Canadians might have concerns about the companies handling their personal data. Burt said Amazon Web Services Canada and Microsoft Canada provide data storage that the government’s AI tools require. But while Amazon and Microsoft are U.S. companies, all the relevant data centres are located in Canada, he said.

Burt said the Treasury Board will shortly release a more detailed implementation plan, which will lay out individual federal departments’ plans, along with timelines.

Dobbs, the economist, said the true value of the strategy will become clear as those details emerge. But for now, “The plan itself says that it is a plan for a plan.”

This article was made possible by Google and the generosity of readers like you. Donate today.