Prime Minister Mark Carney announces a new federal AI agenda, next to Artificial Intelligence Minister Evan Solomon, in Toronto on Thursday.Cole Burston/Reuters
Canada’s much anticipated AI strategy was announced last week, after a months-long delay. The Liberals evidently used that time to pad their 50-page strategy with platitudes, instead of coming up with a framework that addresses the three big questions swirling around artificial intelligence technology.
The most pressing of those is safety: what legal and technological guardrails are needed to keep AI from, for instance, acquiring cyber-offensive capabilities? Then there is the core economic regulatory challenge of fostering AI innovation in Canada. And, following on that, there is the third question of how to manage a transition that will disrupt the economy, eliminate not just jobs, but categories of jobs, even as new types of work are created.
Facing this scale of a challenge, the Liberal government delivered what is primarily a feel-good economic document mired in small-potatoes thinking. And despite the strategy’s ostensible aim to achieve “AI for all,” it fails to meaningfully assuage those who are not all for AI.
Canada ranks near the bottom on that score among peer countries, amid serious and understandable concerns about mass job losses that could result from AI’s productivity gains.
Opinion: Canada’s AI strategy won’t build necessary trust
Canadians who might have been wondering what the government will do when, not if, the AI revolution upturns Canada’s labour market, have no additional clarity on that matter today. There are no details of any safeguards beyond a commitment to “update the laws and standards.” A nebulous promise that AI will deliver 250,000 “good, high-paying jobs” in the next five years does not address the fact that approximately 60 per cent of Canadian employees could be highly exposed to AI-related disruption, and the strategy does not acknowledge the likely job losses except obliquely in mentions of reskilling programs.
A more comprehensive look at the possible downsides of AI was offered recently by Pope Leo XIV, whose encyclical this space will examine tomorrow.
At the heart of the Canadian strategy is the all-too-familiar Liberal approach of spending hundreds of millions of borrowed money on corporate subsidies rather than coming up with a sound regulatory approach and letting the private sector take the lead (and the expense). There is $500-million so that the Business Development Bank of Canada can underwrite small businesses who want to incorporate AI into their operations, $500-million for a fund to support early-stage Canadian AI companies and a $700-million expansion of a fund to improve companies’ computing power.
The strategy’s only meaningful target is about adoption, too: seeing the percentage of Canadian businesses using AI tools go to 60 per cent in 2034 from 12 per cent today. (Amidst those proposals is a smaller, $50-million injection into the Canadian AI Safety Institute, money that it will need to fulfill its new heavy mandate of tracking emerging risks and evaluating AI models.)
Faced with cutting-edge technology that could upend society at large, the Liberals elected to use their standby, shopworn tools: the heavy hand of industrial policy, new and rehashed spending, and piecemeal changes at the legislative edges. “This strategy reads like an extensive Christmas list of money and aspiration without any reconsideration of why the same strategies over the last decade did not work,” Jim Balsillie, the former co-CEO of Research in Motion, told The Globe and Mail.
Canada is not alone in struggling to navigate citizens’ trust issues while pursuing the prosperity promised by widespread AI adoption. For instance, the Great American AI Act proposes requiring developers to publish transparency reports and declare “critical safety incidents” to a body within the Commerce Department, signalling something of a return to Biden-era transparency and safety obligations.
On the other hand, Argentine President Javier Milei last week proposed a framework for allowing fully autonomous “non-human corporations” run purely by AI to operate. Mr. Milei’s plan shrugs at safety concerns, particularly in the context of his sweeping promise to “keep AI unregulated.” But he is at least showing what it looks like when a government thinks about the right questions.
As Canada’s strategy itself acknowledges, “AI raises hard questions about job security, privacy, sustainability, sovereignty, and trust.” If only the government offered some hint at what it thinks the answers might be.