When his company introduced an AI tool to assist its sales agents in 2024, Devin Marsh, 36, never expected he would one day end up training it to replace himself.
Beginning in 2018, Marsh worked as a senior agent handling customer support and sales for telecom giant Rogers through a third-party company, Foundever.
“[Management] was like, ‘Okay, this is a nice tool that you can use to make your job easier’,” recalled Marsh, whom Canadian Affairs agreed to not identify by his real name due to concerns about legal ramifications from his former employer.
“Then slowly it became, ‘Why aren’t you using this more?’ Now it’s a requirement.”
In June of last year, Foundever began laying off employees; within four months, it had let go of more than 1,000 of them.
Marsh, who lives in Oshawa, Ont., but worked remotely, was let go in late September.
“It felt like the rug was being pulled out from under me,” he said.
Customer service jobs such as Marsh’s are among the most vulnerable to AI displacement. But they are far from the only sector being affected by automation.
Some sources say Canada’s policy response so far has been inadequate.
“Our AI policy in Canada is really being focused primarily on the priority of stimulating the industry in Canada … with almost no attention to the impact on work and preparing workers to navigate the changing nature of work,” said Chris Roberts, director of the Social & Economic Policy department at the Canadian Labour Congress.
“That’s a real frustration.”
AI job impact
AI adoption in Canadian workplaces has been lower than in the U.S. and largely concentrated in bigger firms. It can also be difficult to isolate the effects of AI during a period of significant economic disruption for Canada.
“It’s very hard to disentangle the incredible disruption that’s going on in the Canadian economy generally from these longer-term technological changes that are also beginning to ripple through the economy more slowly,” said Roberts.
Statistics Canada research also suggests AI has not yet caused broad job losses.
However, employment growth has been weaker for younger and less-educated workers from 2022 to 2025 — the period when tools such as ChatGPT became widely available.
Experts have also said mass adoption of AI risks making job prospects harder for these workers, who are most likely to perform the routine, rules-based tasks that AI is best able to complete.
“We’re already seeing fewer positions [in clerical and administrative work], because a lot of those tasks are being complemented by AI,” Tricia Williams, research director at the research hub Future Skills Centre, recently told Canadian Affairs.
Other AI-exposed jobs include customer service, junior content and marketing roles, routine digital work, and professional and scientific services.
In September, Ottawa’s Chief Data Officer Stephen Burt said AI adoption will lead to job cuts in the federal public service.
Looking ahead, the World Economic Forum’s 2025 Future of Jobs Report predicts that, by 2030, 40 per cent of employers expect to cut roles where AI can handle routine tasks.
The report forecasts labour market shifts that will create and eliminate jobs equal to 22 per cent of today’s workforce. The report projects 92-million job displacements, to be offset by 170-million new positions.
Canada’s response
Sources say Ottawa has so far prioritized building Canada’s AI industry over managing the labour disruptions it may cause.
That focus was on display in February, when Evan Solomon, Canada’s minister of artificial intelligence and digital innovation, testified before the House of Commons Standing Committee on Science and Research.
Solomon repeatedly framed AI as a driver of efficiency, productivity and economic growth. He emphasized the need to commercialize Canadian research and keep intellectual property from flowing abroad.
“AI is meant to serve people, not the other way around,” Solomon told the committee. He said that the government’s priority is to build domestic computing capacity and help Canadian firms scale new technologies at home.
The government has created an AI Compute Access Fund, which subsidizes access to high-performance computing so Canadian researchers and start-ups can train and deploy AI models domestically.
“Every time that there is an investment, for example, in our public Compute Access Fund, there is a diligence process to make sure that these companies qualify, that they are spending it in a way that will help their business and promote innovation,” Solomon said.
But when MPs pressed him on whether these investments would translate into job losses, the minister did not provide clear answers.
Opposition MPs also noted the lack of a clear timeline for legislation governing AI use and its impact on workers.
These exchanges reflect a broader gap in Canada’s AI strategy, says Roberts, of the Canadian Labour Congress.
“As far as AI policy goes, or even digital transformation of work … I don’t think we’ve seen focused attention to the impacts on the world of work from the federal government,” he said.
In his view, ignoring how AI affects work is “holding back the effective adoption of AI” in ways that could augment and enrich jobs; instead, employers are steered toward using it mainly to replace workers.
During the committee meeting, Solomon said the Carney government plans to release an AI strategy later this quarter. The government has not yet said if the policy will address worker displacement.
Stifling cushions
Others say governments should not be focused on cushioning workers through an AI transition.
Former U.S. senator Phil Gramm, who is today a senior fellow at the American Enterprise Institute think tank, says society only captures the gains from technological change when workers and capital move quickly from obsolete jobs to new ones.
“Society gains from the creative part of the wave when a new technology arises, but loses from the destructive part of the wave,” Gramm, a one-time Republican presidential candidate, told Canadian Affairs in an interview.
“To the people experiencing the destructive part of the wave, it appears to be a negative factor … but from the point of view of the economy as a whole, society gains when the destructive part of the wave frees labour and capital … to go into higher and better uses.”
Gramm says U.S. programs that aimed to support workers during periods of upheaval slowed re-employment and failed to raise long-term earnings.
He pointed to the Trade Adjustment Assistance, created in 1962 to help workers displaced by rising global trade, and unemployment insurance introduced during the Great Depression, as two examples.
Income supports, he says, can reduce job-search intensity unless tightly linked to work requirements and direct job matching.
But Wes McEnany, director of Labor Policy and Coalition Building at the Future of Life Institute, disagrees.
“I don’t think AI is a traditional technology,” he told Canadian Affairs. The Washington-based institute, which also has offices in California, London and Brussels, is focused on ensuring AI widely benefits society.
“This isn’t like instituting the factory line or the cotton gin. This is going to touch every single industry, and it’s happening with such speed that I don’t think this idea that we’re going to just create an abundance of new jobs is realistic.”
When asked what North American countries plan to do about job displacement, McEnany said he hears few concrete answers.
“The response is often, ‘Well, we’ll figure it out,’ or ‘We’ll have to institute some form of [Universal Basic Income],’ or that there’ll be this invisible hand that creates new jobs,” he said.
“There’s no actual plan.”
Europe, by contrast, has taken steps to protect workers.
The European Union’s 2024 Artificial Intelligence Act regulates high-risk AI systems before they reshape workplaces. Employers that use AI in hiring, job evaluations or layoffs are required to inform workers, ensure human oversight and monitor risks. This limits fully automated decision-making.
EU countries like Germany have gone further. It requires employers to consult employee representatives when AI systems affect hiring, promotion or scheduling decisions. The country also offers publicly funded training vouchers for upskilling.
Gramm argues such protections come at a cost. “It is difficult to create jobs in Europe because it’s difficult to fire people in Europe,” he said.
Back in Oshawa, it didn’t take long for Marsh to find himself out of a job. And he quickly discovered that AI had not only put him out of a job, but had also reshaped the job search process too.
Often, AI tools were the first to screen his applications. In one online AI assessment, his empathy was flagged as a weakness, he says.
Fortunately, a human recruiter recognized it as a strength.
“[The human recruiter] heard me talk about my experience and my passion for customer service, ensuring that people are heard,” he said.
By late 2025, Marsh had got a new job as a technical support agent, this time in the Apple ecosystem.
Leave a comment