This week Prime Minister Mark Carney announced MP Evan Solomon as Canada’s first “minister for artificial intelligence (AI) and digital innovation.” There is no doubt that the prime minister is keen on AI, dedicating an entire plank in the Liberal election platform to various AI investments and incentives. He also must be keenly aware that Canada’s tech elite have increasingly spun into the orbit of the Conservatives, frustrated by what they perceived as neglect from the former Trudeau government, as well as their disdain for the capital gains tax.
Part of the prime minister’s plans for supercharging Canada’s economy via AI includes a pledge to balance “the federal government’s operating budget by capping the public service and improving public sector productivity.”
No doubt part of this attempt to improve productivity and manage a diminished public service will be via AI. The Liberal platform makes this clear: “We will be relentless in looking for ways to make government more efficient. The potential of AI to improve services and delivery must be included in that work.”
AI is already present in government, so this is not an either-or question, it’s about the extent to which we want to automate essential public services that Canadians rely on.
Elon Musk’s Department of Government Efficiency (DOGE) has been one of the biggest proponents of expanding the use of AI throughout the American government and, so far, the results are disturbing.
Less than half a year into their creation, DOGE has been accused of using AI within government to surveil the conduct of federal employees, particularly for behaviour or work that may contradict President Trump’s agenda. They have used AI to expedite immigration enforcement and deportations by integrating sensitive data from multiple federal agencies and they have replaced fired workers at the General Services Administration with a chatbot.
But most concerning is what will be done with all the public data that DOGE now has access to.
It’s no secret that data is the most lucrative currency in the tech sector. As Alison Stanger argues, for AI developers, “government databases represent something akin to finding the Holy Grail.” Government databases provide “verified records of actual human behavior across entire populations.”
As Stanger explains, this “isn’t merely more data—it’s fundamentally different data. Social media posts and web browsing histories show curated or intended behaviors, but government databases capture real decisions and their consequences.”
Not to mention that government data also tracks people who might not have a significant presence online, a treasure trove of information that tech companies simply couldn’t access via traditional means.
Allowing a private company to train its AI on government data would give it an enormous advantage over competitors, as well as giving it predictive power that would be unrivalled.
As Stanger states, concentrating that kind of power in the hands of a single private entity would give it unprecedented power to influence society—including electoral outcomes.
Canada will face the same concerns as it expands its use of AI. Should it rely on the private sector? There is no doubt that Canada’s tech elite are salivating at the chance to access this kind of data. What safeguards will be put in place to ensure the privacy of our most intimate information as citizens?
Finally, there is the question of how we relate—as citizens—to our public services when those services are mediated by artificial intelligence. The more AI is expanded, the more autonomy and discretion will be taken away from our civil servants, as decision-making is handed over to an algorithm.
While civil servants have a wealth of knowledge and experience to navigate the system and advocate on behalf of a client, AI does not. This could severely impact the public trust if Canadians thought that important decisions like whether or not you were eligible for a public benefit was entirely decided by a glorified chatbot.
Which raises a larger question: to whom does one appeal to when an algorithm is making such decisions?
Lastly, we know that AI often inherits the biases of its creators, like AI hiring tools, which disproportionately select white males over others. Vulnerable minority groups, who already may have experienced discrimination in accessing government services, may find these biases continue with AI, further eroding whatever trust they may have remaining in government.
For all these reasons, it is imperative that Canada does not replicate the reckless expansion of AI technologies in government that the United States has witnessed.
Moreover, it will be incumbent on the government to address the very real concerns of stakeholders before proceeding with mass implementation. Otherwise what looks like a panacea today could end up further alienating Canadians from their government tomorrow.
