Bank of Canada: AI Will Lift Productivity Without Big Job Losses in 2026

On May 13, 2026, Bank of Canada External Deputy Governor Michelle Alexopoulos delivered a notably optimistic message about AI and the labor market. Speaking at the Ottawa Economics Association and Canadian Association for Business Economics Spring Policy Conference, she argued that artificial intelligence will lift Canada's flagging productivity without producing mass layoffs. Her line, picked up across Bloomberg and BNN Bloomberg coverage, was that "AI will mostly transform jobs — not eliminate them." That framing cuts directly against the AI-layoff wave narrative dominating U.S. headlines and against the pessimism in recent Gallup surveys. For U.S. job seekers reading every doom story as a personal warning, the central bank's view is worth weighing on its own terms.

What did the Bank of Canada actually say about AI and jobs?

Alexopoulos delivered her first major speech as External Deputy Governor and used it to push back gently on the dominant narrative that AI is already gutting payrolls. The Bank of Canada's position, as she laid it out, is that the evidence so far does not show widespread worker displacement from AI. The technology is improving output per hour worked, she argued, and could lift wages, lower business costs, and create savings for consumers. That is a meaningfully different starting point from the framing common in U.S. tech coverage, where each round of tech-sector layoffs gets read as proof that the machines are winning.

The speech sat alongside the bank's parallel staff analysis, which incorporates only limited AI productivity gains into current projections. In other words, the central bank is bullish on direction but cautious on speed. New technologies rarely reshape economies overnight, Alexopoulos noted. They take years, sometimes decades, to filter through industries before the gains show up cleanly in national productivity data. That slow-rollout view is consistent with what Nobel economist Daron Acemoglu has flagged about the gap between AI hype cycles and measurable economic output.

The bank is not claiming AI is harmless to every worker. Alexopoulos was explicit that younger workers in entry-level customer service roles and coding jobs face the most acute exposure. That detail matters because it lines up with what the data is already showing in the U.S., where IT unemployment recently hit 3.8%, the highest since the pandemic. The contrarian piece is not that AI never displaces anyone. It is that the aggregate jobs picture is holding up better than the most-shared LinkedIn posts suggest.

It is worth flagging the political backdrop too. Canada has spent years debating what Governor Tiff Macklem publicly called the country's "Achilles' heel": labor productivity. According to a Statistics Canada analysis released in 2026, the Canada-U.S. productivity gap has widened by roughly 26% since 2000, with Canadian output per hour now sitting at around 71% of U.S. levels. That is the gap Alexopoulos is hoping AI can begin to close. Her optimism is not abstract — it is tied to a specific national problem the bank has been sounding alarms about for the better part of a decade.

How much AI is Canadian business actually using?

The numbers are smaller than the press cycle implies. About 12% of Canadian firms were using AI in 2025 according to figures Alexopoulos cited, drawn from Statistics Canada survey data. That is up from roughly 3% in 2022, a four-fold jump in three years, but it still means the vast majority of Canadian employers were not running production AI workflows last year. Adoption is also lumpy by sector. Financial services, information technology, and health care are leading. Most of the rest of the economy is still in pilot mode or watching.

That distribution helps explain a tension U.S. workers feel daily. Tech and finance dominate business news, so AI rollouts in those sectors feel universal. But the broader economy looks much more like the Canadian picture: a small share of firms running real deployments, a much larger share experimenting, and a long tail still using AI mostly as a productivity tool for individual employees rather than as a workforce replacement. That is consistent with U.S. reporting that 74% of enterprises have rolled back AI customer-service agents and that many 2026 AI rollouts are stalling on broken foundations.

There is a labor-side detail buried in the sector picture that job seekers should not miss. In financial services, survey respondents told the Bank of Canada that AI was improving productivity primarily by automating routine tasks and freeing workers to focus on higher-value work. In health care, AI tools were freeing nurses and doctors from documentation and administrative drudgery so they could spend more time with patients — the same note-taking software use case Alexopoulos called out in her speech. That is a very different shape than "AI replaced my team." It maps closer to job redesign than to headcount cuts, which echoes how Duolingo walked back its AI-first hiring memo once the practical limits became clear.

The monetary policy implication is quietly significant. If the Bank of Canada believes AI will lift potential output without crushing employment, that gives the bank more room to keep rates accommodative when productivity gains finally materialize. Higher productivity expands what the economy can produce without generating wage-price spirals, which means the bank does not need to choke off growth to control inflation. That is a friendlier setup for hiring than the alternative scenario where AI is treated as a deflationary labor shock the bank has to lean against. Investors picked up on the distinction: Canadian rate futures barely moved after the speech, which suggests markets read Alexopoulos as confirming the bank's existing dovish lean rather than signaling a new policy turn.

What does this mean for U.S. job seekers right now?

The Bank of Canada speaks for Canada, not for the United States. But North American labor markets are deeply linked, and central bank framing tends to move in similar directions. The American picture in May 2026 looks structurally similar to what Alexopoulos described, with one key difference. U.S. productivity growth cooled in Q1 2026, and hiring has settled into what we have been calling a frozen labor market: few layoffs, but also few openings. Jobless claims remain low too. The Canadian view that AI is not currently driving mass displacement is consistent with what U.S. data is actually showing, even if it does not match the loudest stories.

The implication for job seekers is not that AI worry is wrong. It is that the worry needs better targeting. Aggregate employment is holding, but exposure is uneven. Entry-level customer service, junior coding, and routine clerical work are taking heavier hits. Clerical work, in particular, is concentrated among women, which makes the displacement question a gendered one too. Meanwhile, healthcare job growth held up in April 2026, and the sectors actually adding workers look closer to the Canadian picture: care economy, technical operations, and senior roles where AI augments rather than replaces.

Federal Reserve Chair Jerome Powell has been making a strikingly similar case from the U.S. side. At Harvard in March 2026, Powell told students that large language models are "an amplifier, not a threat" for workers who learn to use them well, and he predicted AI's near-term bite would land on middle management and back-office roles rather than across the broad workforce. He also flagged the awkward combination the U.S. is living through now — weak job creation paired with a still-low unemployment rate — which is exactly the frozen-market shape Alexopoulos described in Canada. Two of the most influential central bankers in North America are converging on the same read: aggregate employment is fine, the productivity story is real but slow, and the displacement that does happen is narrow and skill-specific. That convergence matters because monetary policy, hiring sentiment, and corporate guidance all tend to move with central bank framing.

The international comparison sharpens the point further. OECD productivity data has shown for years that AI-driven gains depend less on whether firms buy the technology and more on whether they redesign work around it — training, process change, and managerial buy-in matter more than software licenses. Canada, the U.S., and most OECD economies are early in that redesign cycle. That is the gap Alexopoulos is asking everyone to be patient about, and it is also the gap that explains why the loudest U.S. AI rollouts keep stalling: companies are buying tools without rewiring how the work gets done.

Is the optimistic case actually credible?

There are real reasons to be skeptical of any too-clean "AI helps everyone" story, and Alexopoulos herself was careful not to make that claim. Three reality checks belong on the table. First, the productivity boost has not yet shown up convincingly in national data. The bank itself is only baking in limited gains. Second, surveys still show American workers are unusually pessimistic about the job market, and sentiment shapes hiring decisions. Third, the corporate communications layer is doing real work: companies routinely code AI cuts as "restructuring" or "efficiency," which means the "no mass displacement" reading depends partly on what counts and how it is reported.

That said, the cumulative U.S. data does support a softer reading than the doom feed implies. April 2026 saw a notable jump in announced cuts concentrated in tech and federal jobs, not a broad cross-sector wave. Big AI infrastructure spend is still translating into hiring at the top of the stack. Anthropic just raised at a $900B valuation and is hiring aggressively, and Google is openly recruiting forward-deployed engineers to plant AI inside customer organizations. Even the chip layer is signaling continued workforce demand, with Qualcomm's CEO framing 2026 as "the year of agents" rather than the year of mass layoffs. The real story is closer to what Alexopoulos described: a slow, uneven productivity story that hits some narrow worker groups hard while leaving aggregate employment relatively intact, at least for now.

A useful frame for U.S. job seekers comes from a recent argument that the biggest AI impact may land on minds rather than headcount — meaning the change shows up as job redesign, new task mixes, and shifting skill premiums before it shows up as layoffs. The practical move is to read AI exposure at the task level, not the title level. A marketing analyst role where 60% of the tasks are now AI-augmented is a different job than it was two years ago, even if the headcount is identical. That kind of redesign is what is actually happening across both Canadian and U.S. firms, and it is consistent with both Alexopoulos's "transform not eliminate" framing and Powell's "amplifier not threat" framing.

The bottom line for anyone job-hunting in May 2026: the doom feed is real but selective. The Bank of Canada, the Federal Reserve, and the underlying data are all telling a more careful version of the same story. AI is reshaping work. It is not, yet, ending it. The honest play is to target the narrow exposure zones for risk management, double down on tasks that AI amplifies rather than replaces, and stop reading every layoff headline as a forecast for your particular job.

People Also Asked

Q: Did the Bank of Canada really say AI will not cause job losses?

A: Not exactly. External Deputy Governor Michelle Alexopoulos said on May 13, 2026, that the evidence so far does not show widespread worker displacement from AI, and that aggregate productivity gains can come without significant job losses. Her exact line was that "AI will mostly transform jobs — not eliminate them." She also explicitly flagged that younger workers in entry-level customer service and coding roles face the most exposure. The framing is contrarian, not Pollyannaish.

Q: What percentage of companies are actually using AI?

A: In Canada, about 12% of firms used AI in 2025, up from around 3% in 2022, per Statistics Canada figures cited in the bank's speech. U.S. adoption rates vary by source, but the broader pattern is similar: heavy use in financial services, IT, and health care, much lighter use across the rest of the economy.

Q: Should U.S. job seekers feel better after this speech?

A: Calibrated, yes. The Bank of Canada's view lines up with Fed Chair Jerome Powell's recent framing and with U.S. data showing low jobless claims and a frozen but not collapsing labor market. The honest read is that aggregate employment is holding up, exposure is concentrated in narrow worker groups, and the loudest displacement stories may be more selective than representative.

Stay ahead of the market with The Hire Ground

Central bank views matter because they shape how recessions and recoveries get named, and how policy responds to AI in the labor force. Sign up for The Hire Ground newsletter to get the weekly read on what the data is actually showing, which sectors are hiring, and where AI is and is not eating jobs — without the panic and without the hype.