Front Burner - Canada’s bet on an AI boom

Episode Date: October 14, 2025

Canada's first ever minister of artificial intelligence, Evan Solomon, is spearheading what he's calling a "30-day sprint" to nail down Canada's AI strategy. The plan? To figure out a government approach to the technology in order to boost the Canadian economy.Today, we wanted to take stock of the state of the industry in Canada, and a closer look at the Liberal government’s strategy. What could it all mean for our jobs, our economy, society, and environment?Murad Hemmadi, a reporter with The Logic, joins the show.We'd love to hear from you! Complete our listener survey here.

Of the seven great nations that make up the G7, it is Canada that imposes the highest taxes on beer. 46% of what Canadians pay for beer is government taxation. When the G7 leaders get together, I bet Canada doesn't brag about that. Enough is enough. Help stop automatic beer tax hikes. Go to hereforbeer.ca and ask yourself, why does the best beer nation have the worst beer taxation? This is a CBC podcast. Hey, everybody, it's Jamie.

I just wanted to take a minute to recognize some of the folks who've been writing into the show. A very special thanks to Brayden, who had a few suggestions for Alberta-related topics for us, like the current teacher's strike and the Alberta Prosperity Project. We appreciate this feedback so much. You can reach a show anytime at Frontburner at cBC.ca. To tell us about the kinds of stories that you want to hear. And make sure that you're following us on your podcasting app of choice. It is the best way to make sure you catch every episode.

Okay, here's the show. How do we make sure that Canadians benefit from AI? That it is AI for everyone reflecting our values, benefiting Canadian workers. So that is Evan Solomon, speaking with David Common earlier this month on CBC's power and politics. Solomon is Canada's first ever

Minister of Artificial Intelligence, and right now he is spearheading what he is calling a 30-day sprint to nail down Canada's AI strategy, figuring out how the government should approach it and how to use it to best boost the Canadian economy. AI is, of course, probably the single biggest driver of economic hype in recent memory. Billions of dollars worldwide have been invested and seemingly

everyone is trying to figure out how to cash in. So today, we wanted to take stock of where the industry is in Canada right now and take a closer look at that strategy that Solomon and the liberal government are working on. What could it all mean for our jobs, economy, society, and environment? For that, I am joined by Marad Hamadi. He's a reporter with The Logic where he covers the ends and outs of this industry. Marad, hey, it is so great to have you back onto Frontburner. Thanks for having me back. Always a pleasure.

I think maybe the most useful place for us to start is to enumerate what we actually have right now. What does the Canadian AI industry look like right now? How does it compare to other countries? Yeah, so most listeners will probably have heard people talk about how great Canadian AI is. And that's because of research. A couple of decades ago, the federal government put a bunch of money into AI research at a time when not many people were doing that. And as a result, we have these clusters of research in Toronto, Montreal,

Edmonton, other places as well, but primarily those three places where graduate students and doctoral students are working on the cutting edge of AI, or they have been for the last 20 or so years. What that has not necessarily translated into at large scale is companies. So there are AI companies, don't get me wrong, in Canada, some pretty significant ones. The likes of Cohere or Coveo or Ada, you know, the minister in particular likes to name some of these names. But most of them are not at the scale or at the level of success yet of some of the companies, particularly in the United States. Why do you think that is that that research acumen has it necessarily translated?

So there's two reasons. One is that some of the principles that made some of those breakthroughs in what are called neural nets. And neural nets, the easiest way to think of them is that they're sort of AI models that work a little bit like we imagine the brain does. Those underpin a lot of what we talk about

as AI today, the systems that power them. The people that made those breakthroughs often went to work for the likes of Google or Microsoft, you know, other Silicon Valley companies. Now, there's some dispute about why that happened, but it's certainly true that in the era when these breakthroughs were happening, call it the early 2010s. Not a lot of Canadian companies were using this technology,

and the U.S. companies saw something in these people and in this technology and hired them up. Got it. There are other factors about tech in Canada in general, involving money and the availability of capital, of talent, of customers. is these are longstanding issues and they apply to AI as well. And what about companies here that are just using AI, not necessarily creating AI? You know, we know one of the big sales pitches for generative AI is how it's going to create

huge value for businesses. And do we know if a lot of companies here are adopting the tech? Yeah, I like the line that it's one of the greatest drivers of economic hype. Translating that into one of the greatest drivers of economic productivity, that's still work in progress. I mean, it's true that every successive new digital technology, whether that's computers or the internet, Canadian businesses have been slower to adopt than their peers around the world. It does not look like this cycle is any different and that AI is also, they're also being slower to adopt AI. There are pockets of significant adoption and there are in places you

might not imagine, places like financial services, you know, areas where there's a lot of data and a lot of tight regulation actually serve pretty well for AI. So it's, differs by industry to industry, but on the whole, businesses around the world are figuring out how to use this technology, and it's certainly true that Canadian businesses aren't doing it any faster than anywhere else. Let's talk about Canadian consumers now. So I imagine pretty much everyone listening will have probably either tried chat GPT at this point or encountered like that AI summary that you get when you do Google search.

But I'm curious if you have a sense of the extent to which actual Canadians are going out of their way to incorporate this technology in their daily lives. And in particular, whether they're spending money on it. I think the first question is easier to answer than the second. There are the two companies that have a consumer-facing chat interface, that's why they use. So that's Open AI Anthropic. They've both put out some research recently, and they find that Canadians are in general less likely than people in the U.S. to be using, say, chat GPT, but more likely than people in a lot of other places. So they're using it at pretty high rates.

The payment is an interesting question in part because some of these companies operate on what's called a prosumer model. So, like, they give away the main product for free, chat GPT. use freely. But then their subscription rates are actually fairly high because they're designed for people that are using them essentially for their jobs. So, you know, if you're just like meal prepping or travel planning on chat GPT,

you're probably never going to hit the limit that requires you to pay $200 a month. But if you're using it for your work day or day every day, you'll probably pay that. And frankly, you're probably expensing that, right? And just give me some examples of what people are using it for. the meal prepping, the travel planning, I get. But just beyond that. I kind of use it like Google

in a way, like a Google search, but I feel like I'm using it in a very rudimentary way compared to other people. I mean, I think a lot of people are using it in that way. You know, I've heard of cases where people are using it for, say, helping their kids with math homework, right? Like, I mean, it's been a long time since I did high school calculus. And imagine trying to help someone when either you didn't do that or you, you know, you've certainly forgotten a lot of it. So that kind of thing. We are seeing more and more reports of people using it as a sort of social lever or like a companion, people using it. Just to kind of bounce thoughts of, sometimes you don't want to call someone.

Certainly if you are having that thought at whatever, 2 a.m. You're not going to call someone to run it through chat GPT as always on. there are more concerning instances of people becoming dependent on feedback from it, and there are questions about the feedback that it provides. But certainly, you know, people are using it as a sort of auxiliary to their human interactions. Where do people generally think this technology is headed next? Like, what is it on the cusp of doing, potentially? Yeah, potentially is a big, is an important word in this conversation,

because there's the AI God version of it, like what people call artificial general intelligence. And like there's a lot of dispute even in the industry about what this term means. But generally speaking, you could think of it as AI that can match and surpass human performance. So that's one direction that it could be going in,

which is to say the bigger these models get, the closer they're able to do the full scope of things that humans are capable of and then figure out things that we can't conceive of because our imaginations are limited. by being human. The other version of it is maybe this technology eventually gets good enough that you can let it run certain tasks or certain parts of the economy. And it's worth emphasizing that we sort of do this already. Like, if you think about a package that you order online,

some algorithm is routing it, like from the place where it's packed to your door automatically. And so we're already relying on AI to some extent for those kinds of things. but automating more parts of the economy or more tasks within jobs. So, like, for example, it books my dentist appointments. Correct. And that's what's called agents. There's a lot of – that is the sort of buzzword of the moment is agents, which, roughly speaking, you can think of it as AI that can do things for you rather than just give you responses.

It's sneaky, underhanded. They don't want us to talk about it. But in Canada, beer tax increases are automatic. They go up automatically, yes. Even though at 46%, Canada already imposes the highest beer taxes of any country in the G7. Don't they realize automatic is not democratic? To help stop it, go to hereforbeer.ca. And ask yourself, why does the best beer nation,

have the worst beer taxation. Why BDC for my business? The timing's right. Everything's in motion. Economy's changing. It's all about automation, AI. So I said to myself, take the plunge. Yes, I need a loan, but I also need a hand

from a partner who's truly working with me, helping me, no matter what comes next. Not later. Now. Get ready for what's next. With BDC, you get financing and advice adapted to your projects. Discover how at BDC.ca.ca slash financing. BDC, financing, advising, know-how.

Another thing that we hear a lot about is data centers, which are these facilities that house the computing infrastructure, the generative AI tools run on. And what do we have here in the way of those types of facilities right now? And maybe more importantly, what is being proposed in terms of future projects? So what we have right now is not actually the kind of data center that all the AI hype is around. We have data centers right now in Canada that serve software and other digital services to consumers and businesses. So, like, if you use software on your computer or you use the Internet, it's definitely run out of a data center somewhere. It may be running out of a data center in Canada. So, like, Microsoft or whatever company you use, the documents you store in the cloud, those are somewhere that kind of distributed, but some of those might be in Canada.

The kinds of data centers that people are talking about that people are proposing, there's two categories. is one is training. So that's when you build the model that powers the application. Those are the gigantic data centers everyone's talking a lot about, you know, the gigawatts of power, billions and billions and billions of dollars. They're generally very large. Nobody actually does that kind of activity in Canada right now for a bunch of different reasons, not least to which is there isn't the capacity to. That is some of what's being talked about, build a gigantic data center so that you can train the next generation of models and the next generation after that. Then there's what's called inference, which is basically when you ask chat GPT a question, it does an operation to get you an

answer. Those data centers can be smaller. They can be less energy intensive. And there's also a lot of talk about building those kinds of things in Canada. Sorry, this is a dumb question, but why would we want them here? I mean, I think that is an excellent question and a question that the government is going to have to try and figure out a concrete answer One reason is the word sovereign gets thrown around a lot in this context. It's something that the Minister Solomon talks about a lot. Canadians want to know that their privacy and their data is protected from deep fakes, that their kids are protected.

That's really critical because unless they trust that, they won't adopt this technology. So that's critical to our AI strategy. And I talked about this, not just AI for everyone, but this notion of digital sovereignty to make sure that our privacy and our data is protected. So on that, I know that was a... One part of the argument here is Canadians have data that we don't want in the hands of another entity or government. So should that data, particularly say government data that's sensitive, should that be processed on Canadian soil, should it be required to be held there? There are already, to be clear, some requirements around this.

There's that argument. The secondary argument is essentially geopolitics is messy and getting increasingly so. say AI becomes what the most optimistic projections say it will be, say it's running large parts of our economy. We want to be sure someone else can't turn that off. And if the processing and the running of that AI is happening in data centers outside our borders, we have less control over whether somebody else turns us off or not. Okay, so let's bring Evan Solomon into this conversation a little bit more. So you mentioned his thoughts around AI sovereignty, but just elaborate on what you're hearing from him right now

and what we've heard from him in terms of his goals and what he thinks that this technology, and presumably the prime minister, think that this technology can do for Canada. Yeah. So the sovereignty question is certainly a big part of it. The other part of it, The other part of this is the productivity question, right? So Canada has a productivity crisis emergency, certainly a significant problem. We've heard various people in positions of authority say that over recent times. Productivity is essentially output poor workers. So, you know, can we make more stuff with the same number of people? And over the years, Canada's answer has generally been no. AI is looked on as if there is a silver bullet to this problem, AI might be it, in the sense that AI allows each worker to do more with the same number of hours in the day by taking away certain things that they have to do.

If the goal of the government is to boost economic output in order to pay for the services that the government provides, you have to find a way to do that. The government has decided that increasing the population via immigration is not going to be the way we do that, and therefore we need something else. A.I. Is that something else? He's recently struck this task force. He's calling this like a 30-day sprint now to create an AI strategy. And just tell me a little bit about who's involved in that and what we might expect to actually come out of it, a little bit more about that. Yeah, so it's 26 people. They are generally people who are involved in the AI industry. Quite a lot of them come from the private sector. So there are people like Jewel Pino, who's a very well-respected researcher now working at Coher, a company that I've mentioned earlier. People like Ajay Agarwal, who's haunted the Creative Destruction Lab, which is a major hub for startups in Canada. It's people like Killarone, who's a professor at McGill, who's been studying some of these issues of technology in the economy and in society. These 26 people are broken up into groups. They have to go away, talk to their networks, and come up with concrete recommendations over the course of this 30-day sprint. The government will take those recommendations as well as input from a public consultation. All of that's due by sort of the end of this month. And they'll put that

together into a revision of their AI strategy, which they're going to announce sometime in December. Yeah. And just either from Solomon herself or maybe anybody on this task force, are we hearing any kind of concerns, any sort of feelings that we need to pump the brakes here? Like there's regulation that needs to come with this. Certainly the polling suggests that the public is desirous of there being regulation in this area. The quick version of this is that in the last parliament, there was an AI law table as part of a privacy bill. that bill never passed through Parliament for a bunch of different reasons. And Minister Solomon has said that he won't be reviving wholesale, the AI part of that bill, but they do plan to do things on privacy, securing consumers' privacy and data, which our private sector privacy law is 20 plus years old. And so they've talked about that. Certainly, there are people who feel like this process is going too quickly. I think the government's position has been essentially that we're in an AI race, and it's not like there hasn't been a lot of talk about what Canada's

position in AI should be. The issues are fairly well discussed. The recommendations were made as part of the last bill, and certainly on the industrial side, there are lots of people with lots of ideas about what we need to do next. Marad, we began this conversation with you telling us about how this country did invest in research and actually made quite a bit of headway only for American companies to then take our people and build their own companies, right? So if the goal here is to try and create this thriving, sovereign AI industry, how are we going to do that? with a country whose entire population is about the same as the state of California

and with the track record that you talked about earlier. I'd be making a lot more money if I could answer that question. It's true that Canada is a relatively small consumer market, so that's 40 million people, but we are actually quite a large economy. And a lot of what you will hear the eye startup say is it would be really nice if the Canadian government and big Canadian companies would buy, Canadian-made deck to do the things that they need to do. So if you are a startup building an AI product in financial services or mining or energy that can make a

company more efficient, it helps if you have a large Canadian company or the Canadian government or someone like that to say, we will give you your first contract. It doesn't have to be a ton of money, but we'll help you prove that this is useful. And what you hear companies say a lot is that in the states, big companies are willing to try out local companies technology. And that helps them scale up. You know, you've heard this prime minister saying Canada is an export economy. We make things here, whether they're physical things or digital things, and we sell them to the world. AI is not going to be any different if it is to be a thriving economy. No one is building an AI company just for Canada. But it would help if Canadian institutions

were willing to help them along their way. You know, we talked about how there's a lot of talk about the potential of this technology about what it could be, what it might do. But are you speaking to people who are concerned that we're putting like too much emphasis on something that just really might not get much better than it already is? So there's the most utopian version of this that AI helps us solve climate change, create more leisure time for everyone gives us basically the world and the society and the economy that at least a lot of people would hope for. It doesn't have to do that to be economically

significant. So we've talked a little bit about the productivity problem in Canada. If you could free up some time from workers, if we could produce things more efficiently, if we could move stuff across the country more efficiently. And I want to be clear, more efficiently can mean with fewer workers, but it can also mean with the same number of workers just doing things quicker. There is a benefit to the economy from that. There is a growth benefit. There's a sort of growth dividend that that produces. I think the argument that, sure, AI as it is right now, is limited. But you are seeing real outcomes that are beneficial to individual companies and to individual people. Now, is that worth the billions and billions of dollars that are going into data centers

and into AI companies? So that's a difficult question to answer. Is the government making a bad bet by focusing on this? You'd have to assume that it doesn't get any better. It stops today and we never use it for any more than we use it for today in order for it not to be worth a thing paid to put some effort into. And what about the argument that it's such a high bubble right now? I was looking at this Bloomberg piece the other day illustrating all these deals. I'm sure you've seen it that these AI companies have with each other. And it's just like kind of the circular buying of debt from each other. It's just shocking amounts of money. Like what happens to the Canadian economy if and when this hype bubble pops if we're very wrapped up in it, especially? The counterpoint of what we

talking about at the beginning with Canada not having a ton of major companies in this space right now, is that we're not actually as implicated in the bubble as we might otherwise be. Yes. It might actually help us to be kind of small here. Could this be like 2008 for us? We just sort of skirt by. Yeah. We get to skip the bubble. I think the more data centers that go in, the more liability we have on this, for sure. During the dot-com bubble, there were companies that jumped up out of nowhere. They had dot com in the name. They raised a bunch of money. They went

kaput. There's also companies like Amazon that were founded in that era that survived and of current environments. But a thing that a lot of people aren't aware of is this thing called dark fiber, which is that telecom companies built a lot of infrastructure in anticipation that the internet was going to require a lot more bandwidth. Let's build out all these fiber networks because this thing is coming. It's for sure. And so we're going to need this. And a lot of those companies went bankrupt because it didn't happen as quickly as people expected. But all of that dark fiber is being used today. And we're having to build more at a huge pace. The internet expands to fill the available room. If companies in Canada

build a bunch of data centers and the AI bubble pops, those companies might go bankrupt. But that data center capacity will probably end up getting used. I do just want to know if people are talking about the environmental concerns here. MEDA seems to be planning to help to build a data center near Edmonton, which would be entirely powered by its own natural gas power plant. And also, what about Canadian jobs? A lot of businesses talk about this tech being able to automate stuff, right? So if businesses adopt this on scale, what could it mean for Canadian workers? And is anybody concerned about that in the government right now? Do they have a plan?

Yeah, so let's take those in turn. The environmental concerns are real. They are worth discussing. They're generally down to water and power. So data centers use water to cool. There have been advances in how that's happening. So it's not a problem, but there is work happening on that. There's also work happening at places like Mila, the AI Institute in Montreal,

to make the use of what's called compute, which is the output of data centers. You can think of it as like the processing power. to make compute more efficient. But certainly if the new big data center proposals, particularly in Alberta, the idea is to have captive natural gas plants or a natural gas plant that just serves that facility, and they might be connected to the grid for backup power. The best argument that the government can make there is that that is going to happen somewhere, so we might as well capture the economic gains, but that's not a solution to the problem

of the emissions that that produces. In this task force, there is some talk about sustainability, but it's certainly not, you know, issue number one. Issue number one is like adoption and commercialization. On the jobs front, there is a genuine dispute right now about whether the effect of AI on jobs will be to remove jobs or to change the nature of those jobs. So it may be that AI does not automate an entire job, i.e. all of the work that a person does, but automates a lot of tasks within that job. Therefore, either that person has to do a lot more to justify their continued employment or they have to figure out new tasks that will fit in. I think it is notable that you hear companies whose job it is to sell AI to people

talking about how devastating this is going to be for jobs. And another way you could read that is, our technology is so good that you won't need workers anymore. Just to illustrate the point of the government paying attention to it, in last year's budget, so under Prime Minister Justin Trudeau, So there was $2.4 billion dollars put into AI. Two billion of that went into compute, the thing I was just talking about processing power. I believe 50 million went to skills and training.

That doesn't need to be damning. There are the parts of the government that focus on these things, but that's just a sense of scale. I do think there's a really difficult question that people need to answer about late career workers. There's actually not a ton of evidence that suggests that full-scale retraining of workers, i.e. a worker who used to do something before and we're, you know, moving them completely to do something else, that that is particularly effective. A lot of workers, particularly mid and late

career who are displaced from jobs by layoffs, have a very hard time getting back into the workforce, regardless of what training is offered to them and what government supports are offered to them. That's a problem that the government has to be alive to. If it's going to go out there and say we need to use AI to make the economy more productive. Yeah. Maraud, I could keep going. But I think that that's a good place for us to leave it today, though I hope that you'll come back soon. Any time. All right, that's all for today. I'm Jamie Poisson. Thanks so much for listening. Talk to you tomorrow.

For more CBC podcasts, go to cBC.ca.ca slash podcasts.