Canada has world-class AI research and talent but needs to translate these advantages into Canadian-made AI products adopted by domestic companies, experts said during a webinar organized by Mitacs and presented by the Canadian Science Policy Centre.

But the country currently lacks the ecosystem and policies to incentive public and private sector procurement of Canadian AI technologies, they said.

Unless graduates trained in Canadian university AI programs can tackle meaningful problems and develop strategic AI applications here, they will seek opportunities in the U.S. and other countries, the panelists warned.

“The critical element, the recipe for success here is having really challenging, interesting, compelling problems to work on,” said Elissa Strome (photo at left), executive director of the Pan-Canadian Artificial Intelligence Strategy at CIFAR.

CIFAR’s data shows that the Canada CIFAR AI Chairs are the third-highest impact AI research cluster in the world, behind only Google and Germany’s Max Planck Institute for Informatics, she said.

CIFAR currently has 150 Canada CIFAR Chairs, with 2,200 graduates students training across three national AI institutes: Vector in Toronto, Mila in Montreal and Amii in Edmonton.

However, “We know that to keep talent engaged in the field, engaged in a sector, engaged in Canada, they need to have really important problems that align their values, give them an intellectual challenge, give them a technical challenge,” Strome said.

This means Canada requires a multi-sectoral approach, she said.

“Those of us in postsecondary education, in research, need to be working collaboratively with industry, with the startup ecosystem, with the public sector as well, to understand and strategize how we can prepare our trainees to take up jobs in AI in Canada and how we can prepare those organizations to receive that talent as well,” Strome said.

Margo Seltzer (photo at right), department co-head of computer science at the University of British Columba, said currently there’s a gap between the AI programs offered in universities and the programs and problems in companies.

“I think figuring out how to bridge those two together and really connect them in a way that will continue to make sure that the young people do know that in order to become senior engineers, there are jobs to help them do that,” she said.

“There is a lot of enthusiasm from our students at all levels about wanting to do something new and start companies and be entrepreneurs, and we need to be able to support that at a very grassroots level,” Seltzer said.

But she said the first thing she hears from students is that they feel they need to go to Silicon Valley to be entrepreneurs and startup founders.

“I think we need to change that messaging. We need them to know that there is an ecosystem in Canada and that they can have an impact here, and perhaps they can have an even bigger impact here,” she said.

Unlike the giant AI firms building huge foundational AI models to generate as much profit as possible, Canada has opportunities to use smaller, simpler models that are more sustainable and have fewer environmental impacts to tackle issues in societally sensitive areas including health care and finance, Seltzer said.

“We can do that with sovereign compute facilities today. We don't have to build huge data centres,” she said.  “And I think it really gives us a sort of end run around what some of the, you know, behemoths in the industry are doing.”

Canadian companies lag in adopting AI and in AI literacy

One of the problems is that Canadian companies are lagging on adopting AI and AI literacy in the country is “exceedingly low,” said Cam Linke (photo at right), CEO of the Alberta Machine Learning Institute (Amii).

“We’re just really bad at investing and being customers of stuff here in the country,” he said. “And if all of your customers are somewhere else and all of your investors are somewhere else, what are the odds that your company isn't also going to get moved somewhere else as well?”

Canada has some structural problems, such as oligopolies, government procurement that has favoured multinationals and a lot of SMEs that don’t have the budgets to deploy AI solutions, Linke said.

“We like to use our government purchasing power to buy from everywhere else, rather than to buy from ourselves,” he said. “So you've got to stack all those on top of each other and you create a situation where likely nobody's going to be adopting from Canadian companies.”

Companies in the U.S. are typically a year ahead of Canadian firms in adopting AI, he noted. “So by the time we even get to start something up here, we're a year behind a company in the U.S. or elsewhere.”

The problem is not getting ideas out of the lab and commercialized and into the market, Linke said. “It's that loop of customers and capital being able to make a ready ecosystem for adoption. And we're not great at that here.”

Chad Cogar (photo at right), vice-president of AI at the Creative Destruction Lab, agreed that the key for Canadian AI companies is being able to secure customers.

“The key thing is relentless customer focus. That really defines what hits, whether a company hits product market fit and then can scale up,” he said.

“And I can't tell you how many times I've seen all these great inventions that just end up sitting on a shelf because what the founders thought it should be applied to was actually something that nobody wanted,” he added.

Cogar said he was part of the early team at a Canadian-founded AI robotics company called Kindred that had about 30 PhDs in machine learning and was scaling up and needed customers in Canada.

“Instead, me and one of the head engineers were going all around the U.S. to all these different small towns meeting with distribution centers for large American companies, because none of the Canadian companies would come to the table,” he said.

“So one of the consistent messages that we're trying to push for policymakers and working with companies is you need to start adopting these tools, and how do we get companies to adopt tools from small companies,” Cogar said.

Cory Janssen (photo at right), co-founder of AltaML, said bluntly: “I don’t want any more government [funding] programs. Just let us compete for a contract, because there's a regulatory capture piece where, frankly, big multinationals are coming in. We're just giving all the money to the Americans, like the full stop.”

“So if we're going to go elbows up, let's actually have the procurement vehicles, let's have the paths to be able to get in along those lines,” he said.

Meeting with customers for AI solutions is essential

Janssen said Canada needs to implement a model that enables domestic AI startups to build intellectual property that stays in Canadian hands, and provides incentives, such as tax breaks for accelerated depreciation of assets, that encourages the private sector to do so.

“But I think if we fall into that oligopoly trap and just say, ‘Government's going to pick the winners,’ it doesn't matter [then] if the government is a procurement vehicle,” because Canada won’t have a thriving AI ecosystem, he said. “We need hundreds, thousands of companies playing in this space.”

Canada’s three national institutes should be well-funded to bring in the best researchers from around the world, Janssen said. But it shouldn’t be these institutes’ responsibility to lead AI commercialization and technology adoption, because that’s the private sector’s job, he said.

Linke pointed out that Canadian AI companies regularly win competitive requests-for-proposals that are external to Canada, “and then they’re not even able to compete on those RFPs in Canada.”

“So this isn’t about [having] the Canadian version of Microsoft Word is just to have something local. It’s actually leveling the playing field so that Canadian companies can compete for Canadian deals, which right now in many cases it doesn't happen.”

Youssef Helwa (photo at right), CEO and co-founder of FluidAI Medical, stressed the importance of Canadian AI companies developing partnerships and working directly with customers to determine their needs and ensure the AI applications have tangible benefits.

Fluid AI worked closely with clinical customers – both in Canada and internationally – in embedding the company’s engineers within clinical and research facilities to understand their workflows, needed and desired solutions.

Fluid AI has worked this way with the world’s leading hospitals such as the Cleveland Clinic, the Mayo Clinic, the Texas Medical Center and the University Health Network in Ontario, Helwa said.

“And sometimes this goes under the radar or it's not recognized, [but] people treat Canadian talent and Canadian AI as a trustworthy place or a trustworthy position to work with and interact with,” he said. “So we always walk in with that level of trust, that level of specialty, which is very well respected in the industry.”

Fluid AI also is closely associated with the University of Waterloo and has had the advantage of hiring their AI graduates right out of university, Helwa said.

“We get to reach out to them, we get to bring them on board. These are people who are top of their class, they want to stay in the city. We have the opportunity to present them with incredible challenges,” he said.

The Toronto-Waterloo ecosystem also provides talent through the national Vector Institute, the University of Toronto, Western University, Hamilton Health Sciences, and the University Health Network.

“All of these are less than an hour away, which we're able to leverage talent. This is a very, very strong ecosystem for us,” Helwa said.

“In general, and this is across sectors, Canada is very, very good at producing incredible talents from a foundational level,” he said.

“To take it to the end, you have to understand what it takes to deploy, to be with the customers, to how do you make money,” he added.

“You have to be offering value. And when you offer value, you have to take everything from A to Z. You can't just be at the research level working in the research lab . . . ultimately as well, sometimes you just need to be with the customer,” Helwa said.

Corporate Canada needs to invest and create more opportunities

The panelists pointed to the federal government’s recently released national AI for All strategy as a good step.

However, Linke said that “unless corporate Canada decides to buck up and take part in this, it doesn't matter what strategy the government comes up with.”

“It doesn't matter how much money they pour into it. We're never going to be able to pour in as much as corporate Canada can and should as capital allocators,” he said.

“So if we want to have from this be a success in the country and we want more opportunities in labs, a very simple thing is just having companies create more opportunities in industry,” including by paying talent equitably with what they are paid in the U.S., he added.

Public-private investment programs in Canada also need to move faster, Linke said. “The typical cycle was nine to 12 months for a project to get approved. The technology changes in nine to 12 months, and your competitor is way ahead of you, but you need to go through three to six months of internal reviews.”

“We can't do these things. We need to be looking at moving quickly, being aligned, and then being able to review and iterate from there to be successful,” he said.

Seltzer pointed out that universities have experts in AI but they also have experts in medicine, mining and other domains, and those two groups of experts need to be integrated to tackle real-world problems. “It's not just about the technology. It's really about the technology in the service of a real problem. And those real problems have real domains,” she said.

“And so embedding both the domain experts and the technology experts with the customer is absolutely crucial,” Seltzer said.

Another barrier in Canada that AI talent faces is the country’s immigration process, Strome said. “It’s incredibly challenging for them to get here in the first place, to get their student visa in the first place, let alone to stay.”

Even when graduates are able to stay in Canada, they’re attracted by much higher salaries and more affordable housing and cost-of-living offered in the U.S.

Seltzer said in Vancouver’s high cost-of-living environment, “It's hard enough for our graduate students to eat while they're in the graduate program, but they actually do want places to live and to meet after as well.”

“I've had multiple students take leaves from their very successful graduate programs to go spend time in the U.S., and it wasn't because they liked what was happening there. It was all about they couldn't afford to stay here and finish their degree,” she said.

Panel moderator Eva Reddington (photo at right), vice-president, policy program development & government relations at Mitacs, said it’s clear that the investments Canada made in AI a decade ago has established one of the world’s leading AI ecosystems, especially in research and talent.

“But leadership and discovery is not the same as leadership and deployment,” she noted.

“The next chapter of Canada's AI journey is about translating those strengths into broader impact, into productivity gains for businesses, globally competitive Canadian companies, improved public services and meaningful benefits for all Canadians.”

The panelists confirmed that Canada’s AI advantage is real, Reddington said. “But sustaining it will require more than world-class research. It requires strong partnerships, faster adoption, successful commercialization, and a continued investment in people and our talent.”

“If there's one message that emerged from [the] discussion, it's that Canada's next chapter in AI won't be defined by what we invent alone,” Reddington said. “It will be defined by what we build, what we deploy and the value we create for Canadians.”

R$