Tech Won't Save Us - Canada’s Government is Rushing AI Adoption w/ Hadrian Mertins-Kirkwood

Episode Date: June 18, 2026

The Canadian government is going all in on AI, without understanding the real effects of the technology. Hadrian Mertins-Kirkwood joins Paris Marx to discuss Mark Carney’s push for large-scale investment in AI, despite being unable to describe how adoption will work, how Canadians will benefit, and what policies will be implemented to mitigate growing risks and harms. Hadrian Mertins-Kirkwood senior researcher and political economist at the Canadian Centre for Policy Alternatives.The podcast is made in partnership with The Nation. Production is by Kyla Hewson. Support the show on Patreon.Also mentioned in this episode:Hadrian wrote about the Canadian government’s new AI strategy.Paris wrote about the AI strategy and the new social media policy.The Canada Revenue Agency chatbot is expensive and giving incorrect information.The Canadian Immigration Department is using AI that hallucinates when reviewing applications.AI tools intended for Ontario doctors are providing incorrect information.Here’s an overview of the federal AI strategy, “AI for All.”Support the show

We need to take into consideration all of these other metrics. It's not just about JDP. It's not just about getting the cheapest, most profitable AI companies. It's like, what is actually serving Canadians? What is actually serving citizens and communities, protecting our environment and so on? These are also important considerations, and they need to be part of our AI strategy. Hello and welcome to Tech Won't Save Us, made in partnership with The Nation magazine. I'm your host, Paris Marks, and this week my guest is Hadrian Merton's Kirkwood.

Hadrian is a senior researcher and political economist at the Canadian Center for Policy Alternation. Obviously, we talk a lot about what happens in the United States on the show with occasional conversations about Europe and some other parts of the world, too. But I figured it was a good time to have a check-in on what is going on in Canada. Mark Carney has been prime minister now for over a year. He has a very different orientation toward the tech industry than his predecessor, Justin Trudeau. And we're finally starting to see some of his own tech policies, or at least his approach to tech policy, be revealed, right? You know, start to get some meat on the bones. Some meat still needs to be added, but we'll see how that goes over time. And I figured it was a good time to have Hadrian on

because Mark Cardi and his AI Minister Evan Solomon recently announced their AI strategy, right? You know, they were starting to put it the idea of how they are going to approach AI, how they see this integrating into the Canadian economy, and the types of ways that they are going to try to promote its use. I think for me, it's very clear that this is a very industry-oriented and industry-focused policy. And in part that is because industry has been very involved in the process of putting it together. But it's also just the fact that this is how Mark Carney thinks about AI and thinks about this technology. For him, it's very much about engagement and adoption and far less about thinking about the harms that can arise from the use of generative

AI and chatbots in particular. And I think one of the things that is important to note about the strategy is that kind of outwardly, a lot of the discussion is, around chatbots and generative AI, right? The notion of adoption is around these more, you know, immediate and recent instantiation of artificial intelligence, right, of AI tech. But as Hadrian explains, there's actually a lot in this policy about the adoption of, you know, artificial intelligence in a more mundane way, artificial intelligence in the means of what it meant a number of years ago, you know, in terms of like, you know, predictive software and things like that, and trying to to think about the different ways that, you know, sectors might be made more efficient through the

rollout of digital technology and AI systems that are not so much chatbots and image generators, but the types of things that we meant, you know, when we talked about AI a number of years ago, you know, algorithmic decision making and the like. So, yeah, I think that this is a very insightful conversation about what is going on in Canada, how the approach of the Canadian government very much mirrors what we're seeing in many other governments as well when it comes to the adoption of AI and kind of the government focus on trying to attract AI and tech investment into Canada, just as we see in the UK and France and in so many other countries. We also talk a bit about what we're already seeing in the rollout and how the idea that

generative AI in particular and AI more broadly is supposed to deliver more efficiency and productivity isn't necessarily reflected in what we have seen so far, particularly with these chatbots, and that there could actually be a lot more drawbacks that come with adopting them that we should be paying a lot more attention to, but that the government doesn't seem to want to talk very much about. So this is a fascinating conversation. I was really happy to have Hadrian on the show. If you do enjoy it, make sure to leave a five-star review on your podcast platform of choice. You can also share the show on social media or with any friends or colleagues who you think would learn from it. And if you do want to support the work that goes into making

tech won't save us every single week, so I can keep having these critical, in-depth conversations with so many experts from different parts of the world on different areas of, you know, expertise when it comes to technology. You can join supporters like Scott in Cambridge, California, and Chad in St. Louis by going to patreon.com slash Tech Won't Save Us, where you can become a supporter as well. Thanks so much and enjoy this week's conversation. Hadrian, welcome to Tech Won't Save Us. Hey, Paris. Thanks for having me. Absolutely. It's great to have you on the show. It's been a while since I've done a Canadian-focused episode, but with the new AI strategy from the federal government, I figured, you know, it's a good time to do it. And, you know, who better to talk to

but you, who, you know, has been looking really closely into these issues. You know, before we get to the AI strategy specifically, I wonder more broadly what you make of how Mark Carney's government has been approaching, you know, tech policy and AI specifically, you know, over the past year or so that he has been in office now. Like, what really stands out to you about their approach to tech? I mean, the most obvious thing, and it was immediately obvious in the way that, that Mark Carney and the federal government talked about AI shortly after getting elected last year, and it was very evident in their new strategy,

which is that they're very ideologically committed to AI adoption. Now, that doesn't mean they're totally out to lunch or that they're wrong, but the way that Mark Carney talks about it is we must adopt AI sort of for its own sake. And the assumption is, there's all sorts of assumptions that will lead to productivity gains and improvements in public services

and all sorts of other stuff, but there's not really a lot of evidence. It's largely taken on faith. And that's also not unique to Carney and Canada, right? We're seeing that narrative all around the world, this assumption that if only we could adopt AI and if we do it quickly and we do it urgently, we'll see all these benefits. So that's definitely been the overarching theme. I think more recently, because when the federal government started talking like that a year

ago, they did get pushback from people saying, well, what about the costs? What about the risks and the harms? And the big theme over the last six months has really been about trust. How can we build trust in AI in Canada, but not trust for its own sake. Trust as it means to an end. If we can only build trust, then surely Canadians and Canadian businesses will want to adopt AI. So those have really been the big themes from Carney. Yeah, I think that's really well said.

And it stood out to me when they were giving the press conference on the release of this AI strategy, you know, last week as we're talking. You know, Mark Carney was asked about his personal views on. on AI. And he was kind of like, you know, trust is important, but I'm focused on engagement and adoption. And I was like, yeah, that kind of sums it all up, right? How his focus on this is really about getting people to use AI because he believes in it. He really does. And it's, funny. I mean, trust is one of those obstacles, frankly, to adoption the way they talk about it. The other obstacle is literacy. So the way the federal government is talking about AI is Canada is

lagging on AI adoption. And if only people knew more about AI and trusted it more, then this adoption gap would disappear and we would catch up and everything would be okay. And I think if you ask both Canadians, those wouldn't be their top two issues. It's not just about lacking literacy in AI or not trusting the software. It's like there's pretty fundamental concerns that remain unaddressed. Yeah, absolutely. And I want to come back to that. But I want to talk a bit about what we've seen for the past year or so to start off.

And I feel like, you know, Mark Carney and his AI minister, Evan Solomon, have been talking a lot about AI, obviously. and the need to adopt it. And they have certainly been, you know, increasingly showing that in the public sector, right, in trying to roll out AI tools within different government departments. What have we been seeing in, you know,

what the public sector and what the government itself has been doing with AI adoption over the past year? So in some respects, it's been a kind of targeted approach. So we know, for example, that the federal government has developed this tool called CanChat, and it's supposed to be the kind of public version of co-pilot or whatever. as internal chatbot. I have lots of friends in the public service.

They talk about can chat as basically being directly or completely inferior to the commercially available models. And so they would rather use, if they're going to use the chatbot, they'd rather use chat GPT. But of course, the idea behind something like can chat is it is proprietary. It is domestically controlled. So it checks those boxes, but it's not necessarily actually helping the public service. But the point is that there's an attempt to build these purpose-built public service

tools. But there's also a kind of broader push. And we see this beyond the government, where the federal government is saying, we just want adoption. We kind of want adoption for its own sake. We want businesses to adopt AI. We want public institutions to adopt AI. We want individuals to adopt AI and try it out and kind of see what works. Let's start experimenting. And I don't actually think that's inherently wrong. We should be experimenting with new technology. But again, it's the sort of ideological belief that it will inherently lead to productivity. And here's where we run into problems because the federal government in its budget last fall really cut the size of the public service. So we're expecting, you know, upwards of 40,000 layoffs from the federal government.

And the way they talk about it is, you know, don't worry because we're adopting AI. And so any, you know, loss in personnel will not necessarily lead to worse services. That is taken on faith. And that's what worries me. It's one thing to say, hey, public servants, let's experiment with AI tools, especially, you know, domestic tools. Maybe we can find some productivity gains, and then potentially we could restructure the government once we've proven that these tools work. But we're seeing the move it the other way. We're saying we assume that the productivity will come. We're going to start laying off people now and hope nothing breaks. And that just strikes me as a very dangerous way to run a government. Yeah, I completely agree.

And especially when we're seeing, you know, these reports from the private sector about how the investments in AI maybe aren't delivering the efficiencies and the productivity gains that, you know, were claimed would come of kind of using these tools, right? Well, totally. And I mean, there was a very famous study that came out of MIT showing that about 95% of of private sector AI pilots are not producing a return on investment. And I think what's important to recognize there is it actually, that's not even an indictment of the technology.

Like, there's lots of problems with the technology. We could talk about, you know, the randomness and hallucination and whatever. But I actually think the more fundamental issue is that we have no best practices and no evidence for how to adopt AI institutionally. So if you just tell every single business to say you have to use AI because it's cutting edge, that is a solution in search of a problem. And so what these firms are finding is, well, we actually don't know what problem this technology is solving.

And so it's costing us a lot of money and it's not doing anything measurable. And so really that's just, that's kind of an institutional failure. It's a cultural failure. It's, again, this assumption that AI inherently is productive without actually having kind of systems or evidence to support that. Yeah. When the government was talking about needing more small and medium-sized businesses to adopt AI, I was kind of thinking to myself, like, okay, what is the goal of this?

Like, what is the outcome that is being sought? Like, are we just going to see more like AI generated ads in like small restaurants and coffee shops and stuff because they need to adopt AI for some reason because the government wants them to? Like, you know, maybe there's a reason that they haven't started using it. It's, but that's such an important. It seems obvious, right? But it's such an important insight.

Because yes, this new strategy from the federal government says we've got about 12% adoption among businesses in Canada, and we want to get it to 60% adoption, a five-fold increase in AI adoption. Just, again, for its own sake. And the question is, all right, there are definitely, like we, you know, in software, in the tech industry, we are seeing lots of, again, with caveats, but useful application of AI, or at least it's actually making a real impact. But then you look at other sectors of the economy, and it's like, well, hold on. And why does adopting an AI chat bot, make my, you know, small pottery business more productive? You know, how does it, how does it make my farm more productive? You know, there's like so many other parts of the economy where it's not obvious how these tools are useful. And again, it doesn't mean that there's no useful application of AI in all kinds of different sectors.

But just to be like, quick, every news chat GPT, like, that's not a strategy. Yeah, absolutely. I saw a journalist at The Washington Post, you know, kind of make a post the other day. and he's been reporting a lot on AI. And he was like, I think it's reasonable to say that, you know, I'm kind of paraphrasing, that it, AI might transform, you know, generative AI might transform coding and how coding is done. But it's probably being like exaggerated how much it's actually going to make an impact in a lot of other sectors where it's trying to be rolled out. And I feel like, you know, when we're seeing these public sector rollouts, we have already seen a number of, you know, blunders and issues, right?

You know, this chat bot that the Canada revenue agency rolled out that was supposed to be giving. Canadians' advice on, you know, filing their taxes and whatnot and was wrong, you know, most of the time with the information that it was providing. There was a story recently in the Toronto Star about, you know, the use of generative AI tools in the immigration system and how it was like hallucinating, making up parts of people's applications and they were being denied as a result of it. So like, we're already seeing, you know, it's, you know, early days of the government's adoption of this technology and we're already seeing a lot of real problems with the outcomes, I guess.

Well, that's right. I mean, another example recently is the Ontario Auditor General was looking at the use of AI scribes in the medical sector in Ontario and finding in many cases, sometimes more than half of cases the scribes are making up details. And that's important when it's medical transcription. They're making up prescriptions. They're making up diagnoses. That's really consequential. So we're again, we're kind of rushing. We're assuming these tools work that they're going to help. And it's not always clear that they are. And it's certainly not clear in many sectors outside of the most, you know, obvious applications like coding. Yeah, absolutely. And so I want to dig into this AI strategy, right? You know, that they have finally rolled out. There was talk of AI legislation under, you know, Justin Trudeau's government before he left. And then that was kind of put on ice for a while as, you know, the new government kind of figured out what they were doing with AI. And as you were already saying, you know, we have already seen kind of a shift in the government, you know, in recognizing that. they need to try to address some of these supposed trust issues with the technology, right? You know, when we had Mark Carney and Evan Solomon laying out this strategy, what really stood

out to you as like, you know, kind of like the big picture things and then we can kind of drill down into into specifics? I think what's really interesting to me, because this was the part that I maybe didn't expect, is how focused the strategy is on what they call sector specific applications. So the whole thing is framed around like key industries. health sciences, energy, transportation, agriculture, manufacturing. How can we use AI to more efficiently allocate fertilizer in fields? How can we use AI to assist with diagnosis of genetic illnesses? How can we use AI to improve energy efficiency on the grid? And I read that. And I think most Canadians,

most people will read that and say, that sounds pretty good. I actually doesn't sound that objectionable to be using what is really just predictive software. If we stop calling it AI and we just say It's like, let's just use predictive software in these different sectors. That sounds pretty good. And what they don't talk about is generative AI. They don't talk about chatbots. The term chat bot shows up one time in the strategy. So this whole thing strikes me as a bit of a bait and switch, right?

Like all the public conversation around AI is what are chatbots doing and to our brains? What is it doing in our schools? What is it doing in our workplaces? What is it doing to our democracy online? What is it doing in terms of requiring this huge build out of data centers? of course, most of the data centers are being built to service generative AI models. So that's what everyone is thinking about, what people are worried about. And then what there's strategies about is something really fundamentally different.

It's all about these sector-specific applications, which are, frankly, pretty inoffensive. It doesn't mean we don't have to also think about how we regulate, especially health data and so on. But it's just, it seems really out of sync with where the public conversation is at around AI right now. That's really interesting, right? because I feel like I'm completely on board with what you're saying, because it feels like on the one hand, as you mentioned, there is this push to get more of these more mundane, more traditional forms of AI that are often not the ones that we're talking about

as having these big trust issues and what have you rolled out in these different sectors. But then that kind of policy or that strategy is then used to justify, say, the build out of these data centers or putting billions of dollars into generative AI, specifically like companies in Canada that are working on generative AI or rolling out chatbots to post-secondary and what have you? So what do you make of like the difference there how, okay, maybe part of this is focus on the sector role, but they're using that to justify, you know, a big push on generative AI specifically in these other areas, I guess. So I think there's two possible interpretations and I'm genuinely not sure.

But they stem from the same problem, which is that this strategy and the way, and this is, again, reflective of how the federal government is talking about AI. The strategy does not define AI. And that's actually a big problem because AI doesn't mean anything or it means everything and nothing, right? Yeah, it's like a huge omission, right? To not be clear. It is because AI is a broad category of technology of automated computer systems that can do

all sorts of different things. There's a part of this strategy that says Canadians need to learn about AI in order to benefit from AI, which, again, on the face of it is kind of a mundane statement. I know, it's fine. But as soon as you're like, Canadians need to learn to use AI chatbots in order to benefit from analytics

in the agricultural sector and crop allocation and whatever, it starts to break down. Like, as soon as you get specific of what AI means, the arguments don't follow. And as you pointed out around data centers, it's a big piece of that, saying look at all these productive applications of AI in things like medicine,

in things like manufacturing. Therefore, we need to build gender of AI data centers, like, that is actually not A to B, right? So there is a bit of a bait and switch here happening. And I said, so A, that could A genuinely be that the government doesn't understand the distinction, which I do worry about. And then the other is that they are being deliberately misleading by conflating two or multiple categories of AI and applications of AI that are actually fundamentally quite different. This is also true when we talk about regulation, because they talk about, we can't afford to smother innovation, right? And that's kind of classic government stuff. We want to

allow innovation. But just because we want innovation in health sciences doesn't mean we can't afford to regulate chatbots for kids, for example. These are totally different applications. And then to say that we can't regulate both because we want innovation in one, it just doesn't make sense. Well, and I feel like you could see that in the way that the announcement even took place, right? Because, you know, Mark Carney and Evan Solomon were at, you know, this hospital in Toronto, a research hospital. And we're talking all about how, you know, They were walking through these demonstrations of how technology was being used to make health care better. And some of that was AI solutions, but a lot of it was like just other technologies that they were using.

And it felt like the way it was being presented was kind of like, you know, this is even a further justification for these generative AI technologies. When it feels very likely that next to nothing that they were using to like make healthcare more efficient was chatbots and things like that. But as you're saying, more, you know, more traditional forms of AI, you know, kind of the interaction between robotics and those, you know, older forms of AI that have been used for a while. Not like bringing chat GPT into healthcare or these AI scribes or whatever, right? But it felt, you know, the way that it was being talked about, it felt kind of like misleading in that, you know, a lot of the policy response and stuff is focused on generative AI. But that's likely not actually what we're talking about and how the hospital is using AI, right? Well, exactly. I will just add, the actual announcement happened on a stage just surrounded by doctors, right?

It's the prime minister surrounded by doctors. Like, that's the picture of AI. I guess they want us to imagine. But that is not the AI that we are experiencing as citizens, as consumers, as businesses, right? This push around these chatbots, which is just like a totally different thing. Yeah, very well said. It really stood out to me, especially hearing you talk about it as well. And, you know, the way that the strategy kind of lays it out.

in those sectoral areas. You know, I want to talk about, you know, there are many things that come up in what you're saying, right? And a big part of this was focused on training in particular, right? And, you know, when you're talking about rolling it out in different sectors, the government is talking about needing to train people up, needing to train young people to use AI and sending them into businesses to, you know, to help kind of AI roll out.

And again, like it felt like, you know, you're talking about how they're trying to get a lot of different sectors using these older forms of AI that have been around for a while, you know, these predictive systems and things like that. But it really felt in that case that the training was really focused on generative AI again, right? Or at least the way it was presented kind of felt that way. So I wonder what you make of how they were talking about the training piece of this. Yeah, this idea of literacy comes up a lot that Canadians lack AI literacy and therefore the government needs to spend a lot of money to build literacy.

Like they want to, you know, educate like a million young people and put, you know, thousands of teaching kits into schools. So like K to 12 education is starting to integrate AI in different ways. And so this like literacy is really important. And I will say, I think digital literacy is super important. Like we should. You know, we are in a digital world. It's important that we understand how it works.

We understand what the risks are. We have, we can exercise control over it. So I actually don't have a problem with digital literacy. I worry that they are framing this not as a like, let's teach people to think about AI and maybe be critical, maybe recognize when it is not appropriate to be using AI or maybe there's real issues with how the industry itself is working and so on. And instead, they framed it very much as like a technical training thing, which is like, we need to give people education so that they can use and develop and build AI tools.

And that's just a very different kind of, to me, that's not literacy. That's like technical training, right? And again, it's like this, the wishy-washiness of the term literacy leads to these problems in the same way that the strategy doesn't define AI and that leads to problems. Is this kind of conceptual vagueness ends up sounding like one thing and meaning something else. Yeah, I think very well said. And, you know, as part of this, they're talking about making sure that all universities and all, you know, post-secondary students have access to large language models.

And, you know, there was a piece of it where they were talking about, I think, training like 90,000 young people to like send them into businesses to like help them roll out AI basically. And yeah, I don't know. It just feels so weird, right? It does, it does feel weird. Again, this is, I've said this a lot, but it feels very ideological. It's just this assumption that AI is the future and therefore we need to lean in 100%

and just prepare all young people for this AI future without thinking about whether it's appropriate in all contexts and whether it's actually a net benefit, right? I mean, on this job front, they talk about creating like a quarter million jobs, which sounds good. Sure, I'd love, you know, it'd be great if we had a quarter million more jobs in Canada. They don't really explain how they got to that number. But they also don't talk about like, are they net jobs? Or are we gaining a quarter million in like data center maintenance and then losing half a million jobs in like really meaningful employment that people don't want to lose, right? So we don't, we just don't know.

They don't provide that evidence. They don't provide those numbers. It's, again, kind of taken on faith that this will be a net positive. for workers, and in particular for young people who are most exposed. Yeah, absolutely. And I saw an interview, you know, after the strategy was released with Evan Solomon, the AI minister, he was on CBC, the public broadcaster. And he was pushed on this question of jobs, right? Because even in the broader narrative, you know, the impact of generative AI technologies on jobs has been, you know, a real big part of the conversation, right,

and what it's going to mean. And he was kind of pushed on how they've talked about 250,000 new jobs. And he was kind of saying, like, you know, this would be potentially, you know, I'm kind of paraphrasing, but in like white collar industries, in like, you know, kind of tech jobs working with AI. And, you know, the person who was interviewing him at the CBC was kind of like, right, but what does it mean for the people losing their jobs? Like who, how many jobs are being lost? Like, who are those people? Like, what does that look like? And he like couldn't provide, he could talk all about all the new jobs that were going to be made. But, you know, there was no number or figures or, you know, kind of estimates that he could provide on what it would mean on the other side of the equation,

right? Yeah. And I mean, in his defense, no one has a good sense of how many jobs are at risk. So to be like, that's, that's fair to say we don't have a good sense of how many jobs are at risk. But that's not a reason to kind of plow ahead and ignore the risk, ignore the risk. Because, I mean, a lot of the valuations of these companies is, because they're not making money now, it's predicated on 10 years from now, them doing a huge amount of intellectual

labor that's currently done by people. So there's certainly the tech industry is banking on automating a huge amount of labor, even if we're not actually seeing it yet. But, but I think this, the idea that we can just be hands off and hope for the best is, is pretty naive. If I could put out my economist hat for a second, there's basically two ways you improve productivity, because productivity in an economic sense is, is output per hour worked, right? One way to do that is you do the same amount of work with fewer people. And that's, that's one way to improve productivity. You know, if you continue to produce 100, you know, widgets, but you only need five people instead of 10, you double your productivity. The other way to do it is to empower your workers to produce more.

It's like you keep the 10 workers and all of a sudden you're producing twice as much. It also doubles your productivity. History would suggest that employers will opt for the first. If we can just reduce headcount and maintain output, that's the choice we're going to take. It's cheaper. It's easier than investing and training our workers and protecting our workers and trying to kind of build this additional capacity. So the federal government is really banking on the second one. They're like, we just assume or hope that private sector adoption of AI will just allow people to do more and become more productive without costing jobs.

But again, that's kind of taken on faith. And history does not suggest that that's going to be the case for many workers, at least not in the short term. And that's one of those other big problems is, you know, a lot of folks in industry will say, well, don't worry. Remember, look at the Industrial Revolution and, you know, most people were agricultural workers and now only 3% are. everyone else found new jobs eventually.

And it's like, yeah, in the aggregate, over decades, people found new jobs. But like individuals who lost their jobs, maybe didn't. Maybe they just lost their job forever, right? And we're just not really grappling with the consequences of that transformation and potentially the scale. And again, I say potentially because we're not seeing yet, we don't know if and when we're going to see kind of mass layoffs, which is the big fear. But as long as that risk remains, we do need to take it seriously.

Yeah. And let's not talk about all the human harm, the human destruction of the industrial revolution and what came along with that, right? And of course, it really downplays, you know, how a lot of the effect I feel like that we're seeing from AI, you know, early on at least, is affecting that kind of entry level piece, right? And so what does that mean for young people in particular, trying to get into the labor market, the job market, you know, trying to get, you know, their kind of foot in the rung of like the ladder? You know, does that make it more difficult for them? And I wanted to go back to something that you were saying before, but before I do that,

you know, what have we seen in kind of like the union response to this? You know, I'm sure that they probably have thoughts on AI generally, but, you know, in particular the way that the government is rolling this out. Are they skeptical of the government strategy here? Are they on side without? What are we seeing there? I think something that's been really interesting about the union response, certainly in Canada, is that the labor unions are a lot less worried about job quantity right now and a lot more

worried about job quality because that's what they're actually experiencing. What a lot of labor unions are experiencing is not mass layoffs, but they're seeing seeing the integration of these AI systems in the workplace, which is increasing surveillance, increasing practices of algorithmic management, which means you have an AI system that might be managing the hiring process or managing the discipline process. And that's just making work worse for workers. It's eroding their dignity at work, their autonomy and so on. So that's actually the number one concern for the labor unions right now. It's like, can we get more control over AI deployments in the workplace? But over the long run, absolutely they're thinking about the automation risk.

And that's where the public conversation is at. And it's certainly what unions are thinking about it. They're just not seeing it yet, this idea of mass automation. Yeah, that makes sense too, right? Because, of course, we've seen for a while now how these technologies have been rolled out in order to try to kind of claw back the rights that workers have won over many decades, right, by giving more power to management. One of the most egregious examples, just to highlight it, is the Teamsters in Canada.

I've got Teamsters all over the place. They represent a lot of workers in transportation and logistics. And one of their big concerns is putting inward-facing cameras in truck cabs. So drivers, again, this is not an issue, to invoke the Luddites. I mean, it was not an issue of technology in principle because drivers were happy to have outward-facing cameras because it protected the drivers. If they, you know, if someone cuts them off, the driver could say, it wasn't my fault. Look at the video footage.

But now what's happening is employers are putting cameras in the cab facing the driver and tying them to these AI systems where if the driver looks away from the road at all, like if they look down to pick up their water bottle, the AI system says, oh, that's dangerous driving, flashes a light, makes a sound, and reports it back to the employer that this was distracted driving. And the driver's like, I was just getting a drink of water. Like, that's actually not safer. than having me look at the road unblinking for three hours at a time, right?

So this is a big fight that the teamsters are having right now in Canada and the U.S. But I think it's a good example of what do we mean by AI eroding job quality. Like, yeah, people are worried about self-driving trucks. But right now, what workers are dealing with is surveillance, is this algorithm management in the workplace. Absolutely. And I feel like we've seen versions of that in like, you know, long distance trucking and stuff for some time too, right? You know, the movement of technology and they are in order to, you know, kind of take away some of the rights and autonomy that they had in doing their work.

But I wanted to go back to what you were saying on literacy, right, on this notion of AI literacy. Because I wanted to tie it back to something that you were saying earlier on, because it really does feel like the government is framing this around, like, the opposition that we see to AI right now is because Canadians don't understand it, right? And that if we just train them, if we make them more literate on AI, then we'll have more acceptance. And I wonder what you make of this framing and, like, their choice.

specifically to try to frame it that way and to try to make this the way that they, you know, are trying to get, I guess, broader public acceptance of generative AI in particular. Yeah, it is really curious to me. And again, it's like, what does the literacy mean? Because it's not that Canadians are not using AI tools, right? Like, you know, adoption of chatbots and day-to-day life is high in Canada like it is in many parts of the world, even though skepticism is also high. And we see this, especially in Canada and the U.S., which is interesting, more so than a lot of other countries, where use of AI is high and also concern about AI is very high. So people are using these tools and also not happy about them. And they don't think it bodes well for the future

of society and the economy and so on. And so I guess the federal government sees an opportunity here. If we can just get people to believe more in these things, then they'll use them more and that will bring benefits. But I'm not clear on like, what is AI literacy supposed to teach? You know, if I'm in the federal government AI classroom, what are they teaching me to do? Are they teaching me prompt engineering? Are they teaching me about like the inherent limitations of an LLM? Like, it's actually not clear to me what that's going to look like. One thing we know, and again, the challenge with all this AI stuff, especially genera of

AI stuff, is we don't have like longitudinal evidence, right? As a researcher, I want good evidence. And right now a lot of it's speculative. We have kind of early evidence. But one of the big issues with AI, of course, is evaluating outputs. and the idea that you can train someone to evaluate the outputs of an LLM is very central because that overcomes one of the inherent limitations. And the problem is we're really bad at teaching critical thinking, A, and B, you can't

just like shortcut expertise. Like the people who are using LMs most effectively have whatever, a decade, two decades of experience in their field and so they can identify the problems in the outputs. But to take like a recent grad and just be like, use AI to produce your. engineering report and just double check it. It's like that's a solution or that's a recipe for disaster, right? That's not a solution. So I'm kind of rambling here because I'm looking for something to hold on to.

I don't know what literacy is supposed to look like, really, in practice. I know what it looks like in principle, but I don't know what it looks like in practice and how it's actually going to solve the problems the government's talking about. And it's also just like, I don't know, do they expect that they're going to be able to train Canadians to pick out every time a Google AI overview has incorrect information in it? you know, like there's a lot of kind of thinking and checking that needs to go into something like that that I think the average person not only is not doing, but I also think like shouldn't really be expected to have to do just to know if what they're interacting with

is accurate or not, right? Totally. And I mean, this is a bigger conversation about digital media for sure because we have similar problems with social media, right? Historically, we trust that our newspaper editors are, you know, our TV editors to do the fact checking for us. so we could just read it and assume that's all correct. And that has, we've drifted away from that model increasingly over time.

And so it's not that these AI search summaries, for example, are even novel in that sense, other than they create perhaps an unearned sense of authority. And when you see your Gemini summary at the top of your search results, you're like, that's got to be true, right? Because the AI said it. And if it's not true, how would it? I know? Because the reason I'm searching this question is because I'm not an expert in it. And so I have no way to evaluate this output. Yeah, there's just kind of some really inherent problems

with how our information is being curated. Yeah, I was saying to someone recently, like, I notice small errors in it all the time. But it's like, you know, if you're generally approaching it with the trust that used to be given to the Google search engine and what it was producing, then I don't think you would be looking to say, oh, it got this season of a TV show, wrong or it said that it's from you know I put in a quote from a certain Lord of the Rings book and it said it was from fellowship of the ring but it was actually from return of the king right and it's like it's like small things but you really need to check to see if it's accurate or not and you know I don't think the expectation should be there for every Canadian to have to or

every person in generally using these things to have to be like so aggressive in checking you know the answers that that come from from the chat pots from the tools that you know they're not only running into that that it's not just tech companies kind of like putting out into the world, but the government is kind of encouraging them to use, right? Totally. And we've seen lots of cases of that with like these AI power chatbots being used by like airlines and even, you know, governments like the tax, tax agencies and stuff.

And people coming with genuine questions like, can I get a refund or like, how does this thing work? And chat bot just giving them wrong answers and then people acting on that information. And then that introduces all kinds of questions around liability and stuff like that. Yeah, it's interesting because it's, it's a, again, a question of like, is that actually productive, right? Is the fact that it produced output quickly without a person, but then created a whole bunch of other problems,

a net improvement to our lives or not? And I think that question has not been definitively answered in many contexts. Absolutely. I completely agree. And, you know, I wanted to ask you, because you mentioned this trust piece as well earlier, right? And this strategy started to lay out kind of an idea or a framework of the types of regulations that they're going to be bringing forward on AI and generative AI and chatbots specifically. I wonder what you made of, you know, the kind of regulatory approach that they were laying out there and the types of things that they were focusing on and, you know, on the other's hand,

then choosing not to focus on. Yeah. So to the government's credit, they do lay out a lot of the concerns people have. And they talked about like sexual deepfakes, for example, which has been a real concern publicly recently, about, you know, capturing people's data without their consent or knowledge. And they talk about how Canadians should have a kind of fundamental right to privacy. And they talk about funding various institutes and oversight bodies that are keeping track of these risks and harms. So that stuff all sounds good. I think the problem is, there's two problems. One is that it kind of

stops there. So they say, like, we are going to address privacy. Don't worry. But I worry when you're not specific. What does it mean? to protect Canadian's privacy in the context of AI. At the same time that the government's advancing other policies, like a potential ban on social media for people under 16, which would require some form of age verification. So whether or not you agree with that policy, that is anti-privacy as an kind of inherent approach.

But now they're also talking about a fundamental right to privacy in the context of AI. And when you're not specific, it's hard to know how these things are going to fit together. So that's the first issue in the regulatory approach. It's just that it's vague. But the second is that it just ignores some stuff completely. So one issue that comes to mind is copyright.

Lots of folks are really worried about, especially generative AI, that's trained on copyrighted works, on private data, without consent of creators, without compensating creators, and so on. And this is a big debate. Canada has just undergone the kind of copyright review because we're trying to do like, how are we going to deal with this? And this new AI strategy doesn't mention copyright a single time. The word copyright doesn't show up.

one time. So what are we doing with that then? How are we engaging with copyright? The environmental concerns of AI are not addressed a single time here. They talk about we're building data centers and it talks about how Canada has a mostly clean electricity grid. And so there's kind of an implication that data centers are consistent with clean energy. But that's not actually the case in practice. And the plan doesn't make any claim that new data centers are going to be clean. most are being built with gas fire generators. So, and then there's also, I guess, another big category is issues around cognition and mental health. Areas where we, you know, evidence is still early.

We can't say definitively that AI use is kind of ruining our ability to think. But there's certainly anecdotal evidence. There's early evidence that that's the case. It's certainly something we should be keeping an eye on, but none of that is acknowledged on the strategy. So the regulatory approach here is both vague and also incomplete. And that worries me because these are real risks and harms. And we don't have to panic necessarily, but we should be trying to get out in front of them in the same way that we should be getting out in front of potential job losses. Yeah, I agree with everything that you just said that, right?

And on the point around, you know, the effects that we see on, you know, critical thinking on people's mental health potentially from the use of these chatbots, you know, kind of the early research and reporting that we've had. And then to see part of the goal be to roll out chatbots to like every post-secondary student or whatever. It feels a bit, feels a bit odd to be, you know, taking that policy and not thinking through what the implications should be of it. And just to pick up on what you're saying about the data centers, it's like we're seeing a lot more pushback within Canada to these major data center projects, particularly in British Columbia and in Alberta. And yet, as you're saying, the government really does not seem to want to grapple with this, you know, a recent post. Polling suggested 68% of Canadians wouldn't want an AI data center near their community, and they also want strict regulation of AI, even if it slows innovation or development or what have you, right? And so Canadians really don't seem to be on side with this.

And the vast majority of new data centers being planned in Canada are being planned in Alberta, which, as you say, means they're going to be powered by gas, not by renewables by any stretch of the imagination. And also being built in in water stressed areas and being built in the backyards of people who haven't consented or were there inadequate consultations and all this stuff. And yeah, there is a certain, I mean, and I will say, I'm not categorically opposed to data centers. We use data centers every day for everything we do with our devices. So we do need infrastructure to manage a digital economy. That's not inherently a problem.

And I also think a data center is not inherently a problem because you can power with clean power. You can do it efficiently. the issue right now is this is this kind of frenzied buildout of data centers. We're building data centers that are orders of magnitude larger than anything before. We're doing it so fast that we don't have the grid capacity to support it. We're doing it so fast that we're kind of trampling over the rights of communities to consult and so on. And not thinking long term about things like water use, which I think water use is more of an issue in some areas like Alberta, which is more drought prone than others, but still a legitimate concern. And so I think a lot of these problems are a function of the speed.

You know, it's not like people are in Vancouver are protesting every new warehouse that goes up in the city, right? It's that these data centers being so quickly and without consultation and also because they are kind of symbolic points of resistance. I think a lot of the opposition to the data centers is not only about the literal effects of the data center, but the fact that this is where the kind of rubber hits the road of AI and it gives people something to focus on. Yeah, I completely agree. with you, right? The speed and the scale is a big factor in all of this. And I feel like, you know, even listening to what you're saying there, I often feel like the Canadian approach to AI is almost like a mix of what we see in the UK and France. In the UK where they're trying to put a policy

framework together that's going to attract AI companies and AI investment. And in France, where they're really trying to like build out a domestic AI industry, right, and to make sure that, you know, they are one of the key poles of AI development, have their own models, all this kind of stuff. And this was another really key part of the strategy that Mark Carney was rolling out, was making sure to be investing in Canadian companies, in making sure that they're Canadian data centers, that there's a Canadian large language model.

He's very clearly saying that Canada is only one of four countries with this. So what did you make of that aspect of the strategy and the plan to try to invest in Canadian companies, Canadian capacities, and so much of that being focused, it seemed on on kind of making sure that generative AI was something being done in Canada and not just abroad. So I do think this is probably the greatest strength of the strategy. I read a lot of government strategies.

This government loves to have strategies about every single sector. And usually they're pretty frustrating because they're very nonspecific. And to the government's credit, this is a very well-articulated industrial strategy for the AI industry. And by that, I mean it's clear about the vision. It's clear about what we're trying to achieve. It's clear about establishing kind of measurable targets and objectives. And it's talking about, you know, productive interventions by the government at every stage of the supply chain,

making sure we have the compute capacity, making sure we have the R&D capacity, commercialization capacity, like the VC, the venture capital support companies and so on. So they've really thought this through as an industrial strategy for building. a domestic AI industry. So that's, you know, to the government's credit on that front, I think the big question that hangs over it is, is it going to work, but is it even possible for it to work? One of the big struggles is that we're, you know, we look at the big, the AI giants in the U.S., the Anthropics, the Open AIs, with their trillion dollar valuations. The biggest AI-only company in Canada is cohere. It has a $7 billion valuation. So we're just

orders of magnitude smaller. Now, that doesn't mean it's not worthwhile, right? Like, we don't need to have our own open AI to have a viable domestic AI industry. And you mentioned France, and France has taken that approach to. It's like, we don't need to have a trillion-dollar AI company. We can have a like $10 billion AI company that achieves most of what we need to do. And so I think, you know, we're moving in that direction. But there's no grappling here really with the pull of the U.S. industry. of the U.S. market, which has always been a challenge for Canadian innovators and Canadian tech companies where, yeah, Canada does a great job of producing like AI experts and AI startups. And then they all just disappear to the U.S. because the market's a hundred times bigger.

You know, they acknowledge that in this strategy. They're trying to take steps to keep more of that talent, more of that investment in Canada. I hope it works, but I'm not sold. Yeah, I think that's, I think that makes a lot of sense. And I feel like for me, you know, I have a couple further questions on this. I feel like one of my concerns is, you know, I completely welcome, you know, I'm a big proponent of digital sovereignty, right? And I completely welcome the investments in the domestic tech industry to try to build that out to think about what infrastructure is going to look like.

But I'm, I'm like a bit worried that we're putting so many eggs into like the generative AI cart. And like, what if that is something that, you know, obviously there's a lot of hype around this technology? It feels like there's kind of like an economic bubble around the technology. what if we're just kind of chasing what Silicon Valley has set out, you know, the future of technology as being? And that doesn't actually end up being the case. And that we've like put all this investment behind something that maybe isn't going to yield the results that we hoped.

I feel like I'm a bit worried about that. Yeah, I think that's a good insight. And I think coming back kind of full circle, though, when we started talking about the strategy, I think the federal government correctly identifies these particular sectors as good opportunities. It's like maybe we can have Canadian startups that like really name. like predictive fertilizer usage in the agricultural sector, and we build a business around that that becomes globally valuable. I think that's smart, right? Like, I don't think we're, I don't, you know, I think all the support of cohere specifically, like federal government is really

interested cohere, which is our domestic open AI and trying to make it this, you know, viable competitor to the big U.S. generative AI companies. And that just strikes me as doomed to fail. We just don't have the capital. We have, we don't have the head start. It doesn't mean we should abandon cohere, you know, whatever. That company should try and succeed. Maybe it needs some federal support. But like the real opportunities from an industrial strategy perspective are these sectoral things. Maybe there are specific applications in the health sciences, specific applications in energy efficiency or whatever. And I think that's actually where the real opportunity lies. And it is also the like the kind of quiet, uninteresting side of the AI boom. This is more like

traditional machine learning with, you know, building on specific expertise in industrial capacity in Canada. I'm like, that's a lot. That's where we need to focus. And I think that's actually largely what the strategy is about. But it is all, as we've discussed, it's all tied up in the generative AI hype bubble. And that makes it confusing. You know, when the federal government's like, we're going to invest $2 billion in AI, is it like, are we going to be doing it in these potentially, you know,

useful sector-specific applications? Or are we just going to throw that a coherent and be like, try and scale up to beat anthropic? And I'm like, that's never going to work. That's fascinating. And I feel like if that's the way it's focused, it makes a lot more sense than just going after, you know, having our own generative AI company sort of a thing. But since you're an economist, I wanted to ask you another, you know, kind of facet of this, I guess. I feel like there has been a lot of discussion in Canada, just as in, you know, Europe in the past number of years about how, you know, our economy isn't as dynamic or as productive as in the United States. And, you know, the U.S. stock market is taking off so much. And it feels like that's often because these major tech companies, you know, these major tech companies, are located in the United States, right? And we kind of buy so much from them. You know, they operate in our market and many of those profits go back to the United States. Like, is there an economic cost of the degree of dependence that we have, I guess, developed over a number of years on these U.S. tech companies rather than having more of that, you know, kind of innovation and stuff happening

domestically and having more of that work being done within Canada rather than it being done by US companies or us buying from US companies? So it's a great question. It's actually a very complicated economic question. And economists will give you all sorts of different answers to it. Certainly, best case scenario is you do all of the innovation domestically, all the commercialization domestically, and then you sell those goods and services to the world, which is what the U.S. industry is doing. That's kind of best case scenario. But if other countries are going to make this investment, it may actually be beneficial to us to just become dependent on it. And if I can just switch

topics a little bit to this is my other expertise is climate policy. A good example is solar panels. China has invested so much in solar panels that they're like as cheap as paper, right? And it's thanks to Chinese industrial policy that like the entire world can reap the benefits of solar power. That's just so cheap. Every country around the world basically except for oil companies like oil companies, oil countries that are also oil companies like Canada who don't really care about solar. But the point is it's like we might as well piggyback on all of that Chinese investment in solar. it would be foolish to try and build our own domestic solar industry to compete with them at this point. And there is something to be said for that around digital services as well,

that these companies are sinking hundreds of billions of dollars into training AI models and building these AI systems, that we probably get more benefit from just using them, even if we're dependent than we would from trying to build them ourselves. From an economic perspective, that's one argument here. But there's a lot of other considerations that are relevant here, because it's not just about GDP. it's like, what about, you know, good jobs in Canada?

What about like the sovereignty that, like the control that comes with having sovereign AI champions using sovereign infrastructure and so on? That it's like if we take a step back from the economics, if I put on my political economist hat, it may actually be worthwhile to really focus on building out our domestic AI capacity, even if it is less economically efficient than just using American goods and services. And I think that that argument is getting more compelling over time as, the U.S. AI industry gets increasingly belligerent, for lack of a better word, has become so powerful that it is now kind of dictating regulation in the U.S. That becomes an increasingly large risk to Canadians, to Canadian businesses,

to Canadian governments. You know, even if Microsoft co-pilot is the best enterprise option, we probably shouldn't be using it in the federal government, right? We should probably be using something else for security reasons, if nothing else. So it is a long answer to your question. And it is complicated, but I think there's a very compelling case to be made that really should be prioritizing domestic development as much as we can. Yeah, that makes a lot of sense.

And it's always surprising to me to see the Europeans talking so much more about, like getting their governments off of, say, Microsoft services and developing something domestically. And it feels like that conversation hasn't really gotten the, you know, I guess the weight behind it or something in Canada yet, you know. Yeah. Well, and this is a longstanding point of comparison where we talk about, about the EU and European countries being more regulated,

you know, higher taxation, and therefore it's assumed to be kind of hostile to innovation. And on some metrics, that's true, you know, that productivity in the U.S. is higher than it is in Europe or it is in Canada. But other metrics matter too, right? If we look at actual well-being, people in Europe are far better off than Americans on average. People in Canada are better off than Americans on average, even if the richest in the U.S. happen to be doing very well. And so we need to take into consideration all of these other metrics.

It's not just about JDP. It's not just about getting the cheapest, most profitable AI companies. It's like, what is actually serving Canadians? What is actually serving citizens and communities, protecting our environment and so on? These are also important considerations and they need to be part of our AI strategy. And on that front, the AI strategy is called AI for all. Is there any, like, you know, way in to have more kind of democratic say over what this policy is? is it opening it up to Canadians at all?

Or is that not really part of this? It is part of it. They talk about having, for example, sector councils, which is actually an idea I support and we've written about, which is, you know, let's get everyone to the table and figure this out together. Let's get representatives from industry, representatives from labor unions, representatives from government, and try and kind of map out how we want to use this new technology in this sector.

That sort of thing makes a lot of sense to me. Again, my concern is how they frame it in terms of like, let's get all these people together so that we can accelerate adoption of AI in the sector. And to me, that's, now, you're presupposing a conclusion. We should be having these conversations in the same way that we should be having kind of a broad public dialogue. But this evokes something we didn't talk about yet, which was the government's AI consultation in the fall, which was the right idea.

We should have a national conversation about AI. But then what they did was they said it's going to be 30 days long and we're going to get 28 people who are all either in industry or like affiliated with industry in some way to determine what like Canada's priorities are around AI. And it's like, well, hold on. You didn't talk to, you know, you didn't talk to labor unions. It was only after a lot of pushback that one labor union was added to that council. But you didn't talk to civil society broadly. You didn't talk to user groups. You didn't talk to teachers. You know, you didn't talk to just everyone else who's affected by AI. That mindset of like AI is all about productivity, mainly for the private

sector to a lesser extent for the public sector and all the other stuff, all the risks and harms are kind of noise, that orientation, I think, is really dangerous and could be self-defeating in the long run. Yeah, that was really my feeling in seeing the strategy as well, that it did feel very influenced by and shaped by industry. And I feel like, you know, that's in part because of the makeup of the consultation maybe, but I think it's also just a product of the way that Mark Carney and the way that Evan Solomon see AI, right? It's, you know, certainly in part, I'm sure, industry influence, but also just that is their orientation on what AI is and how it should be approached. I think that's, that's right. So Mark Carney's a Bay Street guy,

that's our Wall Street, right, who is just, he sees all of this like all public policy through the lens of how can we increase business investment, how can we increase productivity, and therefore, how can we grow the economy to benefit all? I don't think, you know, Mark Carney's being malicious here. He's not trying to pull the, pull the wool over the eyes of Canadians. I think he just, he genuinely sees AI as a way to boost productivity and therefore grow the economy and therefore have the benefits trickle down to the rest of us, which is a very familiar model for public policy over the last 50 years, right? And then the fact that this strategy ignores a lot of these issues we talked about, copyright, environmental concerns, cognitive

concerns. Again, I don't think it's malicious. It's not like, you know, Mark Carney is out to get workers or out to get artists. It's just they're all peripheral to his worldview, which is How can we just grow private sector investment, grow private sector productivity, and all benefits flow from that? And that is, again, ultimately ideological, with unfortunately a lot of potentially collateral damage. Yeah, and I think you're spot on, right? And we see that not just in tech policy and with this AI strategy, but you can very much see it in the environmental policy too, right?

And kind of the focus on building out new fossil fuel infrastructure and projects, because that will help the GDP in the short to medium term, regardless of kind of the, the broader, you know, kind of human and environmental consequences that'll come with that, that doesn't seem to really factor into the equation. Same with the AI strategy. How do we get more tech investment, more AI investment into Canada? And, you know, we'll deal with the consequences of that later if there are any, and we don't think there are many big ones, right? Yeah. And it's, I mean, again, one of the challenges, I've said this several times is just we don't have a good handle on what the consequences are. And we don't,

no one can really forecast the future. You know, no one knows where, we're going to be five years from now. And that makes this just very challenging. And it's hard to know with some of these concerns like, for example, the erosion of entry-level hiring. So we know that it's increasingly making sense for firms to say, you know, instead of hiring, you know, 10 junior analysts will hire three and then make up the difference with AI, you know, law firms. Let's stop hiring junior, you know, legal assistance, academics being maybe we don't need as many research assistants. You know, there's lots of applications where, yeah, maybe these tools aren't good enough to replace experts in their field, but they are good enough to replace a lot of those entry-level roles.

And that does lead to productivity gains in the short term. But then maybe 10 years from now, all of a sudden, it's like, wait, we've replaced all our junior software devs with Clod, and now we have no senior software engineers who can actually build programs. What are we going to do? And I'm not necessarily saying that's the future. I'm saying it's one possible future that we're not really considering. It's this erosion of the workforce, not to mention, of course, all the social costs of not giving young people jobs. The same is true with these concerns around cognition. Are we eroding our critical thinking skills, our human judgment?

And is that going to come back to bite us later? Maybe. But we don't know for sure. And that makes all of this just very precarious right now. Absolutely. And so, you know, this AI strategy is laid out. Where do you think things go from here with the government's approach to AI? And how do you think, you know, Canadians are ultimately going

to respond to something like this, given the, you know, widespread opposition that we see to data centers and AI within the public, is this actually going to win people over? Is there going to be a real kind of budding of heads between the public and what the government and industry want? Well, I think one thing we can say for sure is that the rubber is really starting to hit the road. So, you know, AI is not new. Generative AI is not new. We're kind of three years into that era of kind of public awareness of generative AI and so on. But the federal government has largely been able to just kind of coast on rhetoric around AI. Say, don't worry, we've got regulation coming.

You know, we support adoption and principle. We want to see data center investment, whatever. Well, it's happening now. The data center investment is happening. The protests are starting. The regulation is starting. At the time we record, the government is about to announce potentially restrictions on youth using social media. So, you know, we're getting this regulation on online harms and privacy and so on.

So again, the rubber is starting to hit the road. And that's going to create a lot more points of friction. I will say one thing that is very helpful about having this strategy out is it does give us concrete things to talk about. Because as we've said, it's just hard to talk about AI when AI means everything and nothing. Well, now we have something to hold on to. We can talk about what does AI literacy mean very concretely. What does it mean to have a fundamental right to privacy for Canadian?

What does it mean to have an industrial strategy that wants to build domestic AI champions? These are kind of concrete proposals that we can actually have concrete discussions around. And I think that's where we're going to be now. We're going to move from these kind of abstract conversations around AI and principle and start having these more granular conversations about all the different dimensions of AI that are actually going to affect our lives. Yeah. And it'll be interesting to see how that evolves too.

And, you know, your mention of the kind of online harms legislation moving forward. There's a new regulator that's going to be part of that too. And I'll be curious to see how that's defined and what that is supposed to look like. But Hadrian, this has been really fascinating. You've given us a lot to think about and, you know, a lot of knowledge and insight into what the government is doing here. So I really appreciate you taking the time. Hey, this was a lot of fun. Thanks again for the invitation. Hadrian Mertens Kirkwood is a senior researcher and political economist at the Canadian Center for Policy Alternatives. Tech Won't Save Us is made in partnership with the Nation

magazine and is hosted by me Paris Marks. Production is by Kylie Houston. Tech Won't Save Us relies on the support of listeners like you to keep providing critical perspectives on the tech industry. You can join hundreds of other supporters by going to patreon.com slash Tech Won't Save Us making a pleasure of your own. Thanks for listening and make sure to come back next week.