How can AI truly boost productivity?

"AI is probably our best opportunity to reverse the adverse economic and productivity trends that Canada has been experiencing for half a century," says Joel Blit, associate professor of economics at the University of Waterloo.

"We've tried to do all kinds of tweaks to innovation policy and things like that, and really it has moved the needle relatively little."

That blunt assessment follows a speech from the Bank of Canada addressing the topic of AI and productivity. In a May 13 keynote, deputy governor Michelle Alexopoulos said AI "represents a significant technological advance that has the potential to boost productivity and raise living standards… By lowering costs for businesses and improving efficiencies, AI could support higher wages, reduce prices for consumers and spur new investment.”

The speech offered what might be called muted optimism, citing “small productivity gains” that are starting to appear in the data, AI adoption among Canadian businesses that has quadrupled since 2022, and the bank incorporating “limited gains” into its projections.

"For now, AI adoption remains concentrated in a few sectors," said Alexopolous. "This means that even if some firms are seeing benefits, it will likely take time before significant gains show up in overall productivity data."

But the Bank of Canada expert was also quick to point out that productivity is not about asking people to work harder: “It’s about how efficiently the economy transforms work into goods and services.”

Canada's productivity problem

Canada has lagged the U.S. and many OECD countries on productivity for decades, and the consequences show up in GDP growth, wages and innovation capacity, according to Simon Blanchette, adjunct lecturer in the Desautels Faculty of Management at McGill University.

“Some factors include, for example, population size, relationships to work, but also… access to training,” he says.

"We've been talking about labour shortages and skill shortages for a very long time.”

Blit points to structural factors — an industrial mix weighted toward lower-productivity sectors, relatively small firms that invest cautiously, and a management culture that tends toward risk aversion — as persistent drags.

"Part of it is that we just don't have any superstar firms that are hyper productive,” he says. "There's tons of reasons. It's hard to point to one specific one. It's also hard to address, obviously."

Blit, who is also co-founder and co-director of the Canadian AI Adoption Initiative, says Canada needs to think bigger.

"We need to do something a bit more radical. AI is a technology that's going to fairly radically change the foundations of the economy and of society. And so it gives us an opportunity to really change that economic trajectory that we’ve had for a half a century or so."

Time saved not necessarily productive

One of the more counterintuitive points in both the Bank of Canada speech and expert commentary is that the most commonly cited metric for AI's workplace impact — time saved — does not automatically translate into productivity growth.

A recent survey from Indeed's Hiring Lab found that 57 per cent of Canadians who use AI at work were saving one to two hours a day, and 22 per cent were saving three to five hours. Workers reported using the extra time to take on other tasks, improve quality or invest in professional development. Alexopoulos noted this as evidence that "AI boosted their productivity."

But Blanchette argues the conversion is far from automatic.

"It’s great to save time but… you need to be really purposeful about what you do with the extra time.”

^{Source: Bank of Canada}

Research on small and medium-sized enterprises, which make up 99.8 per cent of Canadian businesses, shows they save on average slightly over one hour per day through AI use, according to Blanchette. But he cautions that the economic benefit depends on what happens next.

"Even if half of that is being reinvested in terms of strategic planning, innovation, productivity and so on — value-added activity — that could be adding nearly $13 billion to GDP. But that goes beyond AI and leads to another [finding] and that is that implementing AI is important, but what is even more important is everything around AI. So, culture, leadership, vision, making sure that you're actually upskilling your talent."

Replacing and reimagining processes

Blit frames the same issue through an adoption lens. His research on past disruptive technologies identifies two distinct stages: "replace," where organizations keep existing processes intact but swap in the new technology, and "reimagine," where they rebuild processes and business models around the technology entirely.

"The first phase is always the easy stuff... The really big productivity increases are when you then start to reimagine the processes and the business models and entire industries around the technology," says Blit, citing Amazon's transformation of book retail as an example.

For Canada to fully reap the benefits of AI, just like previous technology, it must start reimagining processes, business models and industries around the technology, he says.

"At this point, all we've really done is we've kept all of that intact and just sort of superimposed the AI into existing processes.”

Alexopoulos made a similar observation in her speech, drawing a historical parallel: when factories replaced steam engines with electric engines in the early 20th century, productivity gains were modest until manufacturers realized the technology enabled an entirely different factory floor layout — the modern assembly plant.

The measurement gap

Both experts identified a significant gap in how Canadian organizations — and the federal government — are measuring AI's impact.

For example, employers are using the newer technology with existing processes, so everything is done 20 per cent faster, says Blit.

“That’s great, but the big gains are going to be by fundamentally reimagining entire business models and industries. Those in some ways are harder to measure because you have nothing else to compare it to.”

Blanchette describes what he calls a "task diagnosis": breaking jobs down to the task level to identify precisely where time savings are occurring, rather than tracking productivity at the employee or team level.

"The first question is, where did the saved time go? Not just 'Simon is saving time' or 'Sarah is saving time' — where exactly are you saving time? And then from a strategic planning perspective, if we have more time, what could be valuable for them to be doing and invest that time in? [It’s about] planning for the reallocation, instead of trying to say, ‘You have more time, just find something to do.’"

Oftentimes when we talk about job or task redesign, the mistake is it's not just better work, it's just more tasks, he says, “and that's not good for motivation."

^{Source: Bank of Canada}

Blit is critical of government measurement efforts as well.

"We need to have ambitious targets as a country and we need to have the right measures to measure our progress and obstacles and we're really not doing a great job," he says. "What we've seen so far from the government is they… basically said to StatsCan: ‘Here you go, it's now in your lap. By the way, you don't get any additional dollars’ — I think we need to be more ambitious in this regard."

Who's getting it right — and who isn't

Organizations that are successfully converting AI adoption into measurable productivity gains tend to share a common trait, according to Blanchette: they have a plan.

"They start with specific pain points. They find a solution that actually matches what they need… and then they really try to do piloting in specific workflows, assess success if it’s working or not, have metrics in place — for example, how many hours are we actually saving, translating to how much money we're actually saving."

He points to a small Montreal training company called Oxy as an example. After building an AI agent and testing it within a defined workflow with clear KPIs, the firm calculated savings of nearly $12,000 a month — roughly equivalent to a full-time employee.

Blit raises a related concern: even well-intentioned employers may be underestimating the pace of change and risk-averse.

"We're still sort of stuck with, ‘Oh, should we adopt AI assistants?’ Well, that was last year's news. It has now moved on and the risks are now an order of magnitude higher," he says, referring to the shift from AI chat tools to autonomous AI agents that can operate inside processes, access external tools and work without a human in the loop.

"The fact that so many companies didn't even adopt chatbots because they were risky has me worried as to what's going to happen with AI agents,” says Blit.

Cultural and competitive liability

Canada's traditional risk aversion — often cited as a strength in the financial sector — may be working against the country in the current AI moment, according to both experts.

"I would say caution is good on paper, but in practice, AI is progressing really quickly," says Blanchette. "As I like to say, ‘Today is the day that AI will be the least powerful in our lifetime.’ I feel that we're always playing catchup… we need to realize that because AI is not waiting for anyone, we need to step up the pace.”

Blit is equally direct.

"We are a little bit risk-averse; we're a little bit complacent. We don't have nearly as much competition, especially in some key sectors, as in the U.S., and so there aren't the same competitive pressures. When things are cozy, sometimes you don't adopt as quickly as you might."

He also points to public perception as a compounding factor.

"Canadians tend to be very negative on AI… Every time the media reaches out to me, they want to talk about killer robots and the loss of jobs and all the negatives, and rarely do we talk about all the potential opportunities of AI,” he says. “I think we are already falling behind in terms of adoption."

AI literacy: the missing piece

Both experts converge on AI literacy — across the workforce and throughout an individual's career — as the most underdeveloped element of Canada's response.

AI needs to be a civic skill, just like reading and writing, says Blit.

"It'll be integrated into so many of the things that we do. From a national strategy, we need to… segment our population and figure out how we can best reach all the different individuals,” he says, including high school students, university students, the general population, workers, key decision-makers and entrepreneurs.

The AI literacy piece is the one thing that has really not been on the radar nearly enough, says Blit. “And so what does that mean? It means we need to empower every Canadian… over the course of your career and your life."

Blanchette echoes that framing, with particular emphasis on the quality of human participation in AI-assisted work.

"When we talk about AI, we often talk about humans in the loop, but we need skilled and competent humans in that loop with critical thinking and literacy. That's the key aspect. And that goes back to having more ethical AI adoption as well, because we have more educated people."

HR key to AI

Blit sees HR professionals as sitting at the intersection of virtually every challenge AI is creating — from job redesign and talent strategy to employee mental health and organizational culture.

"I think they're going to be right at the centre, potentially even more than the CEO," he says. "They're going to have to deal with the whole human side of this, but also they're going to be dealing with it from a strategic point of view — ‘How do we reorganize? How do we deal with workers that are resistant?’ [Or] the excessive workload because the productivity gains aren't there, the mental health side of it."

For Blanchette, the prescription comes back to the same fundamentals that determine any organizational change initiative: plan deliberately, measure precisely, and treat the technology as a strategic and cultural transformation rather than a product rollout.

"AI, yes, it's a tech change. But, at the same time, it's much more [about] a strategic change, a people change and a culture change. The tech is important, but we need to emphasize all of the ecosystem that comes with it."

“Taking the best of AI and having guardrails in place to minimize the more dystopian scenario — that's going to be a choice that we're making as a society."