What Does Innovation Actually Mean? (talking research, the academy and AI with Joel Blit)

Okay, Paul, asking for a friend, I have a question for you. I would like to know in your words what innovation actually means because it's this core part of CIGI, the Center for International Governance Innovation, and you're the president. So I feel like you know, because I feel like every time I hear the word innovation, someone somewhere, or innovative, someone loses their wings, but I know there's more to it than that.

Hey, good to see you, Vass. Innovation's one of those terms you hear when you hope people don't ask you what it actually means, but it is so important that you've got a... Everyone should have their own way of describing it. I think what I would say in very basic terms is that it's the how do you create or improve an idea or a thing that solves a problem or makes life better, It's so fundamental. And business schools, of course always try to teach this, what's the course for it? And they can't, despite their best efforts in lots of courses. It's a critical creative function that people get out of. I'm sure university and schools help, but it's quite a creative thing and it's as much of an art as a science for sure.

I'm fascinated that for you, innovation necessarily makes life better. I wonder if sometimes innovation can make some things worse for some groups or subsets. And to be honest, I feel like I never really thought about it. For me there's, as you said, an element of iteration and improvement associated with innovation, it connotes an elevation and a newness, and I think a value associated with, a new opportunity for either something to be accomplished and something to be done, but that there is value there. And people are very intoxicated by the promise or the premise of generating and capturing value from innovation. It's what we crave, it's what we push for, it's what we hope for. And I think-

Yeah, it's energizing.

Yeah.

It's a buzzword that everyone likes. And it is a double-edged sword, as you say. It's like technology, what we talk about here on the podcast, there is always a double edge. Even if you think of the positive and people are excited about that, there's going to be, what about this, and what about that? And then at CIGI, the idea was to put that together with governance as one of the most important things for countries in the world to get right. And so we need innovation in that governance space, new ideas about how institutions should evolve or even be replaced, not about global government, but just about institutions and getting governance right, and the international system functioning hopefully in the interests of humanity. Again, that double edge that you're saying. It's a tension.

I like how CIGI puts those words together because I think a lot of people presume that governance is the antithesis of innovation or that they don't go together. And I think the way that the think tank has intentionally presented governance as something that is deserving of and in need of new approaches and new ways of doing things, obviously appeals to me.

Yeah. Another thing in this podcast, we talk about the acceleration of technology, about the adoption, about just things coming at us, and sometimes [inaudible 00:03:42] grab them quickly and sometimes they're sleepers. And so it's accelerating, all of this is accelerating around us and the decision making, the impacts on governance and things are huge.

Absolutely. Today we have the pleasure of speaking with Joel Blit. Joel's a senior CIGI fellow, so he's part of the community, and he's an associate professor of economics at the University of Waterloo where he chairs the Council on Innovation Policy. We're going to speak with him about what innovation is, but also what it isn't and how innovation goals are playing out, including in an academic setting at the university. And of course, your friend and mine, AI is going to come up as part of that conversation and so too will the future of work.

Joel Blit, great to see you and welcome to Policy Prompt.

It's obviously great to be here. Thanks for having me.

I wanted to ask you, and I know it's tough to define, but what is innovation? How should we be thinking about it?

Innovation is very hard to define. So I teach a graduate class on innovation, and the very first thing that we do is try and define innovation. And the best definition that I have is just doing things better, finding ways to do things more efficiently. One definition that I really like is doing more with less. I like that one because it captures both sides of it. So you can innovate to be able to do more, produce better products, more services, higher quality things, or you can also innovate to cut costs, so doing it with less. Doing more with less really captures both of those aspects.

What concretely is happening now in the world of universities writ large or academia in the innovation space? Are they approaching innovation in a different way recently? Is it the same old? What's going on on the innovation question in the university setting?

I think most professors are approaching it in the same way. Most professors are curious about a certain topic, about a certain question, and they're going after that question. It's out of curiosity, it's because they're just interested. But I think for the institution as a whole, there's increasing pressure to make sure that whatever research is happening is being translated to have real-world impacts. And specifically, say for example in Ontario now, there's a very clear mandate to try and commercialize as many of the ideas and to have an impact in the provincial economy.

Clip:

In the last provincial budget, the Ontario government announced its intention to move to a performance-based funding model for colleges and universities. And here to help us understand just what that means and the impact it could have on the education-

There's a bit of tension here where you've got that the individuals that are doing the research at most of these universities just want to play with their fancy shiny toys and do research not necessarily linked to anything in the real world. There's lots of exceptions, of course. But the institution as a whole is being held to a higher standard by the government that at the end of the day is funding a lot of this research. And we're seeing also with a lot of the granting agencies, more and more of the money that they're handing out there has to be some an impact at the end, some partnership with industry, etc., etc.

So I think we are moving away a little bit as a society from basic research and towards more applied research. I think to some extent that makes sense. You get more immediate returns. But at the same time, if no one is doing the basic research, then where are the next big ideas, the next big set of innovations going to come from? To some extent, basic research almost by definition doesn't pay. There is no obvious immediate application. And so if universities are not doing it, then who will?

I'm just going to come back again because something really interesting here for me that when you think about the way... Think of applied innovation, applied research versus more curiosity-driven science innovation through that, it does make you think a little bit about the models in Japan, Korea, South Korea, where they were very focused in the university on an aligned industrial policy that included applied innovation, but they didn't have as many maybe at the breakthroughs at the more fundamental level because they were very focused on the applied. Is that true? And are we heading a little bit to more of an Asian model with the way we're approaching industrial policy and universities potentially?

I certainly think there's pressure to move in that direction. Whether we're going to succeed or not, I am not sure. Whether it's good or not, I'm not sure. To some extent a model makes sense from an economic perspective because others can be doing that basic research. So if the basic research is being done in the US for example, there's very little holding Canadian researchers back from taking that basic research and then applying it in a Canadian context to a commercial problem, etc. And so in a globalized world where whatever basic research you're doing is flowing out of the boundaries of that jurisdiction, you have to ask the question, to what extent does it make sense to invest in a lot of basic research?

Now, the reality is even in our internet world, even in our interconnected world, there is still some degree of localization of knowledge spillovers. So when you do that basic research, more of that knowledge that you've created flows to people that are geographically close to you than to folks that are further away from you. But that localization is maybe not big enough. If the localization isn't huge, then it might not make sense to do all that much basic research.

The other piece in all of this that we also can't ignore is a lot of this research is happening in part to train the next generation of talent. And so for example, if we're no longer doing a lot of this basic research, so our future talent may not have the same opportunities to learn about this and that and different techniques just because it might be different with applied research.

Policy Prompt is produced by the Center for International Governance Innovation. CIGI is a nonpartisan think tank based in Waterloo, Canada with an international network of fellows, experts, and contributors. CIGI tackles the governance challenges and opportunities of data and digital technologies including AI and their impact on the economy, security, democracy, and ultimately our societies. Learn more at cigionline.org.

I want to speak a little bit more about this shift at postsecondary institutions. We're all very lucky to be rooted at different universities in Ontario. You mentioned the word pressure, and I think pressure at the academy comes probably forward in very subtle ways. But let's step back for one second. We were talking about commercializing research. What does that actually mean? What is a metric of success when it comes to commercialization? And maybe it would be important for us to tie that back to this goal or holy grail of innovation.

Yeah, I think that's an excellent question. How do you measure commercialization success? And there's lots of metrics out there. You can look at startups that have been created from research, you can look at number of licensing deals, you can look at the value of licensing deals. You can look at the number of patents, assuming that some of these patents are actually being used, and then you can track those patents and see if they're being renewed, if others are building on that knowledge. You could try and track number of jobs that are being created, that might get a little bit harder. So I think there's quite a few ways you could try and measure. I don't think any of them are perfect.

I think one of the things that we're seeing now is that we're... Again, at the provincial level especially, we're less interested in partnerships and more interested in partnerships with local firms, with Ontario firms. And again, that is putting a certain pressure on universities, especially given that in many of these research sectors, they're very specific in a very specific area. It's not clear that there's any company in Ontario that does that specific thing. And so it is a little bit dangerous.

The other reason why that this is all a little bit dangerous, and this goes back to some of my own research, some of my early research, is that multinationals, one of the things that they do is we can think of them as a bridge for the flow of knowledge. And so I was mentioning earlier that knowledge when it's created, this [inaudible 00:12:57] flows locally to people that are near you geographically. And so let's say that there's a place on earth called Silicon Valley, it's generating a lot of knowledge and we are very far away geographically.

And so how do we get access to that knowledge? Some of my research shows that one of the ways that you do that is you set up a subsidiary in that remote location. But the other way is that a multinational that might have their headquarters there, but also sets a subsidiary here, actually generates knowledge flow from Silicon Valley to other firms here, say, right?

So this branch plant framework that's often criticized maybe is good from a knowledge translation element. Is that what you're saying?

Exactly. Exactly. So I always say on the one hand, yes, and a lot of people like Jim Balsillie have very strong views on this. We've actually clashed on this a few times. But from a knowledge flow perspective, absolutely, it seems that having these players that are in many different locations do have this benefit that they make knowledge flow back and forth. And the last thing we want from a knowledge perspective is to be this isolated island that has very few links to the outside world. And so getting back to your question, one of the things that worries me is if we are cutting all of these partnerships with, say, foreign multinationals, a lot of these formal multinationals know what the most important questions are, have the most high-tech, cutting-edge capabilities, and so they're also contributing that and they're bringing all this knowledge. And so I do fear that it's a little narrow-minded. On the other hand, I sympathize with the idea that Canadians, taxpayers are investing in the research that we're doing, to the extent possible let's make sure that it's benefiting the Canadian economy. So I really see that tension there.

Joel, you and I have both been in the room quite often when you've talked about your framework for in a way the transformation of AI in the economy, which I think is super interesting. And I've seen very different reactions to it, which we should unpack a little bit about... I won't mention where they're from because of CIGI rules, but let's say an Ivy League institution in the US where people in science and technology studies or legal scholars don't see the world this way. But it makes a lot of sense to economists and policymakers often. You've got your framework of the replace, reimagine, recombine, and then describing how AI works through that as other technologies have in the past. Can you say a little bit about that framework and then let's unpack a little bit the criticisms and the strengths of that argument?

Sure, sounds good. So let's be honest, and it is a fairly simple framework, what it's doing is it's drawing on the literature, most of the economic literature on past disruptive technologies and how they have unfolded, how their adoption paths have unfolded over time and summarizing. It's a high-level conceptual framework for helping people think about how the AI transformation is going to play out, and perhaps more importantly, what are going to be the opportunities over time, over the coming decades? And so this framework basically says, "Look, in the first phase..." And again, this applies to general purpose technologies, so things like the steam engine, electricity, computers, and the ICT revolution, and now AI. And so in the first phase, the replace phase, this is where the new technology displaces the old technology within existing processes.

And so the key here is that the processes don't really change, the business models don't really change. All you're doing is you're saying, "Oh look, here I'm doing this particular task with the old technology." And by the way, when I say the old technology, I often mean we're doing it manually, people are manually doing this thing. "With the old technology, and I can actually take the new technology and just drop it in and do it faster, better, more efficiently, etc. "That's the replace phase. And that always brings some benefits. And so in the case of electricity, for example, in the factory floor, we took out the big steam engine and we put in a big electric engine, which was more efficient, more reliable. And so that brought some productivity improvements.

And when's that? You're talking 1890s, 1900, that kind of stage of development?

Yeah, exactly. So this is very late 1800s, beginning of 1900s. Yeah, these factories used to have a big onsite steam engine that was a single power source. And then every workstation was connected to a single power source through these line shafts. And that forced the factory floor to be designed a certain way because these line shafts couldn't cover too long a distance, they couldn't make any turns, etc, etc. So they tend to be very vertical factories. Now, if you fast-forward 20 years, we move into the replace phase. So we now have electricity and people say, "The electric engine, it's fantastic. Let's replace the steam engine with the electric engine." So they created a big electric engine, they put it there, took out the steam engine, but the factory floor remained the same, everything else stayed the same. That was the replace phase.

Then comes along the reimagine phase. And the reimagine phase is where people finally fully wrap their heads around the true potential of the technology and they begin to reimagine processes or business models or even how that the firm is organized. And so in the case of electricity, people realized that, "Hey, wait a second. With the steam engine, it has to be run at scale, but with the electric engine, we can actually have a bunch of little electric engines instead of having a single power source. We can give every worker a little electric engine at the workstation. So now we no longer need a line shaft, and now we no longer need this constraint of having to be connected to a single power source, and we could design the factory floor in a much more logical way."

And specifically we made sure that the outputs from one station became the inputs of the next station. We had the modern production line, so the conveyor belt, etc., etc. And as the best example of this is a Ford Model T factory from exactly that period from around 1915 or so. So these are the first two phases. And then the third phase is where the technology combines with other technology to create entirely new technologies. And if those new technologies are also potentially disruptive general purpose technologies, then the process repeats again. And so for example, electricity combined with advanced materials to create semiconductors and transistors and the computer, etc,, etc.

So that's the framework. Again, it's a very high level framework, but the key is initially we're limited by our imagination and we keep doing everything the way we always used to, just replacing the old technology with the new, but eventually our eyes open up and we start to get much more creative and really drive big changes in business models, processes, etc.

Can we apply this model to the shifts at the academy that we were just talking about in terms of different expectations or how postsecondary institutions are recalibrating to maybe emphasize or focus on the production or pursuit of innovation? Is that a reimagining phase or is there an incompatibility there?

Yeah, we can apply it to any sector. And what I would say is, and we can use the example of universities, but almost any sector. So with universities, we are at the very beginning of the replace phase, and even the replace phase is taking a very long time. So what can universities do with AI, say? AI can help us create content, write emails, sort through emails. These are all fairly low-hanging fruit. And even there, it's not clear that we're doing very much of it. A lot of universities have put very tight controls on whether their employees can use AI or not, never mind the students, the employees, the staff.

So I'd say we're not even doing that easy stuff, that replace kind of stuff, both in the operation side, but also I would argue in the research side. Maybe on the research side, it's happening a little bit more. It's more decentralized. You've got researchers that are experimenting with technology, finding cool ways to use it to do their everyday research. What is really, really not happening, but again, this is not happening in pretty much any sector, is the reimagine. We have an opportunity to reimagine what universities look like around AI. So we can imagine a future... Are you familiar with the 2 sigma question in education?

No. Uh-oh, should I be? No one's ever asked me it.

No one's ever asked you? I hope I'm not getting it wrong. I think it is called 2 sigma and not like 3 sigma, but I think it is-

We can google it.

So the idea here is that if you can provide custom education to students. In other words, instead of providing the same type of education to everyone, you can see how everyone learns differently and you can provide feedback that is custom to them and you can hold their hand and bring them from where they are to where they need to be on an individual basis, you can improve the performance of students by two standard deviations. In other words, every student could be performing at an excellent level. So this is a result from way back when. But up until now, that's been impossible to do. And the reason why it's been impossible to do is it's just way too costly to have a private tutor for every single student. Now we have a technology that has the promise of delivering customized education for every single student.

We should be jumping on this like... What an opportunity. Imagine if we could increase the learning, the performance, the outcomes of students by 2 sigma, just absolutely crazy. And yet I'm seeing very little of this. I'm seeing that Khan Academy, for example, is talking about this and there's a few players, but the institutions, the universities, the colleges, I've seen almost nothing there. To be fair to them, reimagine is very hard. In order to do that, you really have to turn the whole learning machine upside down and reinvent the way that you're teaching. And so it's scary for people and who knows what's going to end up happening, but it's something that we have to do.

And I think there's two sectors specifically where we need to do this. The first one you just mentioned, it's education, and not just universities but also high school, elementary school, etc. The other one is healthcare. And those are two sectors where it's government-funded, where it's all tacit knowledge driven, where it's very labor-intensive and it hasn't changed for decades and decades and decades and decades. And we finally have a technology that has the potential to change these sectors. And we need to, and especially in healthcare. If we don't do it, I fear that we could lose our quality public healthcare.

Two things here that are coming into my head and one is that occasionally when you're reimagining something, you have to be provoked to do so. And it does feel to me that COVID-19 has provoked a lot of reimagination that is partly being enabled by new digital interfaces and hybrid environments and things. So we have potentially unleashed a little bit of a moment that is going to push some things that otherwise have traditional inertia behind them in a way that to resist just as things do. But I wanted to make sure that I raised the criticisms that would be out about such a model, about just the idea that are we taking as a given that technology is relatively neutral somehow with this kind of model, or are we assuming too much about a progression here that's of value, that it's like we can't control this direction, it's value free?

This is the thing that comes out sometimes if we're too technologist, technologist if that's the right way to say it, when we look at these things if we take on your paradigm. Where are the values? Where's all that in this formulation?

So I think you just posed two excellent questions. So I think you're absolutely right that there's a saying never waste a good crisis. A crisis is people get jarred, it's an opportunity to think differently, to get out of the mold, to try something new. As a matter of fact, I wrote a paper in Canadian public policy at the beginning of COVID saying that this was going to do exactly that, it was going to drive innovation and productivity. I was quite certain of that. And I don't think it did. I think it did a tiny little bit, but I think I was mostly wrong. I think people did change a little bit. And then the second that we could go back, everyone just went back. It was just more familiar. Most of us don't like to teach online, going back to talking about universities and teaching. Yes, things changed, but relatively little. I think the momentum, the inertia is just so big that even a major crisis like that I think just had a small impact in the grand scheme of things.

The snapback like an elastic band?

It did. And even our economy as a whole, you can see that GDP, it's like a V-shape, it goes straight down and then straight up and everything keeps going as if nothing had happened.

And then on the values question of it's too agnostic about values, it's too market efficiency-driven as a model to really be something that captures everything we should be talking about. You're maybe not suggesting otherwise, but what's your response to that comment?

I think it's incredibly important that we talk about the impact of technology in society, that we don't just blindly let it do what it's going to do. For example, I've got many colleagues that do that. To some extent, that's where I started on the whole AI question. I was the guy that with CIGI's help six years ago went in front of the G7 representatives, the Sherpas, and told them, "This AI thing is coming, this is going to be the likely impact on jobs" back in a time when no one was talking about it or almost no one. "And we shouldn't just let it wash over us. We should start planning now. And there are things, policy things that we can do." It's funny because at that point, my job was trying to convince people that this was real, it was going to happen and we should be thinking about it. I no longer have to do that, of course. Now it's a complete different challenge.

So I don't have to be convinced. I've always said that we should try and plan. But at the same time, to some extent, some folks I feel have gone way too far in that direction in the sense of saying, "We should just completely stop this thing, stop this thing and then..." I just don't think that's realistic. I don't think that's useful especially given that we live in a globalized world. If we were to say, "Stop this in the west," do you really think that China is going to stop? And so do you really want to create a world where you are basically saying no to the industrial revolution? Obviously the modern industrial revolution and let others take the mantle and lead? I don't think that's the world we want to be in. At the same time though, we want to try and do it as responsibly as possible.

Something you're getting at though is that postsecondary institutions are saying no in some ways to the transformative potential of AI. So we've talked about that commercialization element, but in terms of adoption or embedding or the delivery of material, even the cadence and the pace, I think it's fair to say that these institutions have been remarkably resistant to change during times that are transformative. So as we look ahead, you mentioned planning, as we look ahead to plan for maybe how AI will continue to recalibrate the labor market and shift the skills that are maybe the hottest or most in demand and change jobs, what is the role of postsecondary institutions in supporting and facilitating that planning? Or do you think that they'll just be totally resistant to that and we'll see supplementary learning opportunities start to fill those gaps?

Let me start by saying that I think it's unfair for us to pick too much on universities.

Oh, they can't fight back, yeah.

They can't fight, right. What you just described of institutions being resistant to change applies to pretty much every private firm, every government department, every institution, everything that I have talked to.

Totally agree.

When I go out there and I'm giving an executive course, you know when they pay attention the most? When I talk about the bias problems, the alignment problems, the lack of transparency, then they perk up. And then they say, "So that's why we're not going to be adopting this thing anytime soon." People I think are just naturally a little bit afraid of change. And any excuse not to I think is always a good excuse, so confidentiality, there's a list-

Proprietary information, I always hear that. That's on my bingo card.

You go into a bureaucrat's office in any large organization and there's a list of 20 reasons why you can't do something. You have that discussion and then maybe you start talking about doing something right, but you got a long list to get through, right?

Exactly.

Anywhere.

I think people would now say that I'm a a rah-rah-rah for AI. To some extent, yes, but only because... And again, just to remind you, I started on the other side, but mostly because I see this tremendous resistance. And in Canada especially, we are among the most negative countries in the world around AI in spite of having some of the best policies and governance around AI. So I think it's really important to state that, that overall all of our institutions, everyone is resisting. And by the way, what gives me hope is that this is a fundamentally democratic technology that everyone has access to. So it doesn't have to be top down.

And so for example, to share a story, again, an executive course, the CEO says, "That's why we're not going to be adopting this technology anytime soon." And the marketing lead over in the corner says, "Actually, we've been using this for about a year now." And that's the way that I have some hope. It's going to bubble up. And universities, the same thing. Maybe the central administration says, "This is just too risky," but individual professors and staff are using it every day, and whether you want to or not, they're going to be using it.

On the point you just made, Joel, about who's using AI, the example that I always go to is the moment that search engines, the basic search engine on the internet is totally AI enabled, everyone's going to be using AI unless they stop using the search engines. So it's like people that tell me they won't use AI, it's like good luck because you'll be using it. There's this incremental emergence of AI into the things that is going to surprise people I think that feel they're not going to use it. Perhaps you could ban it, but that goes back to your banning question. But Vass on innovation in university settings, what would you add to the discussion so far that what you're seeing?

I was being a little bit maybe cruel or unfair to the academy. I think postsecondary institutions have evolved to accommodate more part-time learners, more adult learners, people who are working part-time coming back to school. That used to be unheard of. You couldn't access certain scholarships, everything, the dominant form of learning was an immersive full-time student. We're seeing adult learning opportunities through continuing studies. So facilitating lifelong learning and pretty affordable snack-sized, bite-sized learning opportunities that aren't capital C credentialed, but can help boost people in their current role or as they're searching somewhere. I think that's important. I think the academy is still highly relevant there.

Because just as an aside to the many bees in my bonnet, which you can both see that I'm wearing the micro-credential space sometimes feels a little bit scammy to me because, and I worry that some people might be taken advantage of in terms of what they might be paying for, and it's maybe lower quality or the returns aren't there. Long way of saying, I see experimentation at the university. And when it comes to AI, there are capital P policies at these institutions around options you can give students in your classes. And people are having the conversation about how to use and when to use this technology in a way that complements not just learning, but thinking. Yeah, a long and rambly answer.

Thanks, Vass. I want to cover something else on AI, unless, Joel, you wanted to come in. Did you want to say anything on what Vass just said?

Yeah, and sorry, Vass, I feel like I also didn't answer your question. I went off on a different direction. But to answer that question, and what Paul just asked, I do think that universities are recognizing, Vass, as you were just saying, this model of four years and out, four years and now you're an alumni and now we're just going to ask you for money, that's no longer relevant. It really needs to be establishing a lifelong relationship with the individual because we know that the labor market is churning faster and faster and faster and faster, and people are going to have to continuously reinvent themselves and upscale and move into a different area.

And so giving people everything they're ever going to need and then sending them out is just not realistic or productive. And I do think universities realize that and they're moving in that direction 100%. At our University of Waterloo, for example, we have a brand new continuing executive education office that didn't exist before precisely because this is so important. I also don't think that universities see their job, and I don't think it should be their job, to train people for specific jobs. And there's a push now for university-

There is.

Yes.

No, no, go ahead. Yeah, yeah.

There is. And I think that is wrong-headed and unfair. What universities should be doing... Because if universities are training people for specific jobs, they could finish training them for a specific job and then two years out, that job no longer exists and then that human capital is lost. What universities should be doing is giving people foundational skills and transferable skills, things like say entrepreneurial thinking and entrepreneurial skills, problem-solving, critical thinking, writing, and if you want interpersonal skills, all these things that are very broad and transferable. And then if you want in the last year, fine, we might still have a four-year where in the first two or three years you have all that great stuff, in the fourth year, you maybe get trained for a specific job. Knowing that, if that fourth-year stuff is no longer useful, you can fall back on the first three years. But then to some extent, it's going to be on companies and businesses to do that customization for what they need. And we're not seeing that much of that either. Yeah.

When you look at the literature and, Paul, turn to you in a second. Over time in surveys, employers have always said that they felt university graduates were not job-ready on there. That perception hasn't changed no matter how much the academy has so. I always find that a little bit interesting.

Yeah. Totally good point. On AI, I want to make sure we talk for a second about, in a way, the big economist question right now, which is, in a nutshell, what's the impact on productivity, let me say, and therefore how many jobs will be created versus how many will be lost? And whether that's a net loss or a transition question I think is a legitimate thing, but also whether the inequality impacts are going to outweigh the broad society benefits in the sense that it was just going to increase those gaps more. You mentioned access earlier, will people have access to what they need? And economists are all over the map on it, which is... It's quite fun to watch economists battling on these kinds of things. And the data, you can prove different arguments with different data, but how would you weigh in on that debate about how much do we know, do we have to wait and see on these things? How worried are you about access and inequality?

Yeah, so Paul, that's a huge question and really well posed by the way, because usually people ask me, "How many jobs are going to be lost?" And that's where they stop. And what I really like about the way that you pose a question is yes, there's a jobs question, but there's also the inequality question. Definitely there's going to be some job loss in the coming decade or something like that. And I'll mention more about that in a second. As economists, we very often say, "There's going to be some job loss, but it's okay because there's also going to be job creation over the medium to long run and net net everything is going to be great because whatever jobs are lost, new jobs are created." I do think that's true. Barring the AI singularity, it's going to be true with AI as well.

And so I'm not overly worried, except that this transition period can be really difficult for people. if you have say, truck drivers that are being displaced by autonomous trucks, it's not clear how easily they're going to be able to make the transition to other kinds of jobs. And economists have too often been just too quick to dismiss the transitory pain. if we look back at the industrial revolution, today most of us would say it was a great thing. Obviously maybe not the smokestacks and the conditions, whatever, but overall in the long arc of history is a great thing. But for about an entire generation, real wages didn't really go up. And so if you had gone into the workforce exactly at the beginning of the industrial Revolution, you would've said, "This was terrible." Instead of being out on the land, I'm now in this grimy factory, I'm not getting paid anymore. It's unhealthy. And so the transitions really matter. And that's the first point I want to make.

And by the way, the best estimates I've seen are that about one in five jobs are not going to really be impacted by AI. About three in five jobs, they're going to see between 10 and 50% of their tasks being impacted. And about one in five are going to see more than half of their tasks being impacted. So they're going to be very impacted. And those are probably jobs like translators and writers and journalists, to some extent programmers, but again, I think they might just shift a little bit what they do. So I think the jobs issue is absolutely huge.

But the second part of what you asked Paul is, what about inequality? And that's the thing that is not just transitory, that tends to stay in the long run. However, for the first time we have a technology before us that might actually decrease inequality. The last big technologies, we had industrial robotics, and industrial robotics basically displaced factory workers who tended to be towards the bottom of the skill distribution, and therefore it increased inequality in society, income inequality in society. And then we had computers and computers mostly displaced people in the middle of skill distribution, so back office workers and clerks, etc., etc. And those folks either could get upskilled and become white-collar workers or they ended up in service jobs. In either case, it created a polarization, and so it also increased inequality.

But for the first time, we have a technology that it would seem is mostly going to be displacing white-collar workers, or put differently or looked at it from the other side, is it's going to be empowering lower-skilled workers to perform at the level of a higher-skilled worker. So this is something that we haven't really talked about yet, but the way that I think about AI is it's a machine, it's a model, it's a tool for capturing and sharing tacit knowledge. Because what it's basically doing is it is looking at a whole bunch of data that embodies this tacit knowledge. So for example, doctors every day are making diagnoses and they're coming up with treatment plans. And if you look at the data of what actions they've been taken, what their diagnoses were, what actions they're taking, what the outcomes are, you can create a model of what the best action is in any circumstance, in any environment.

Now give this model to a nurse, that nurse can now be empowered with this AI model to perform at the level of a doctor. And so what is going to happen is that we are going to be empowering less well-trained people with less tacit knowledge, with the tacit knowledge of those that have more of it to perform at almost that same level or perhaps even at that same level. This is terrible for family doctors, my wife is one, but this is absolutely fantastic for that nurse and maybe even someday that PSW.

Right. And a nurse, not even necessarily a nurse, the new category of prompt engineer that is out there, of somebody that just knows the right questions to ask and how to dialogue and leverage. But there's so much we could talk about here for sure. I know, Vass, you had some other things you wanted to get on the table for sure. We could keep talking about AI for ages and we'll have to invite you back, Joel, for part two-

For sure.

At some point.

With our eyes on the clock maybe, how do we think about how those gains or leaps can be captured and capitalized on? So that nurse may be performing more, do their wages increase or are we just viewing that as an efficiency gain? And something else that strikes me, Paul, and I don't know if it came to mind for you when you were starting to talk about the inequality aspect is related to the ownership of these technologies and the-

Yeah, the ownership. And the system allowing the benefits to be more broadly shared. It's not even just a does it enable, does it empower people to leverage these technologies and benefit from it, are they really going to get the systems will allow a little bit of a wealth transfer or an income transfer anyway?

I think that's an extremely important point. I often talk about income inequality and how it might actually decrease income inequality, but the great unknown is who is going to own these systems and is everyone going to have access to them? And if the answer is no, then you could get a different form of inequality. Those that have access to AI and are empowered by AI and those that don't.

Clip:

According to the IMF, half of us will benefit from higher productivity, but the other half the rise of AI could see lower salaries, reduced hiring, and even some jobs disappearing altogether.

And that truly could be a catastrophic world if there's the haves and have-nots in terms of access to AI. Completely agree with that. That is a different kind of concern, but incredibly important.

I said at the beginning we were going to talk about trade and we've spent I think less than two seconds on it. So I've got to ask in the context of all this change going on in innovation, industrial policy, AI, your transformation of a general purpose technology, where are we heading on trade? We're heading into a context where we've got to review the new NAFTA within the next two years, in two years time. How do you see this playing out? Are we going to get bullied by the US and just held to their latest whim on trade, or can we position ourselves for that future? And you don't have to answer that exact question if you have a bit of a different cut on it, but where are we heading on this?

The last time around in the trade negotiations, I was very afraid that we were going to get a very bad trade deal for Canada. It seemed that we did not have much bargaining power. Now, I'm not going to say we got a fantastic deal, but it was pretty good given the circumstances. Depending on who is in the White House when we are renegotiating, we could again find ourselves in a very difficult position. And so absolutely I worry. And I think there's lots of win-wins, I think those are not going to be a problem. But there's also lots of areas where it's clear that one party is pushing a different agenda in the other. I think one of those key ones is around IP, intellectual property protection, which again, intangibles is a bigger and bigger part of the economy around data and flows across borders.

I do worry that there could be things put into that next trade agreement that could really hurt Canada as a whole. And to be very specific, I think intellectual property is currently way too strong, and I believe that it's too strong because of certain groups in the US and other big IP-producing countries that are pushing for it. And I think once again, this is an area where Canada's going to have to resist.

Yeah. And they are negotiators really well on IP and these intangible questions that we haven't perhaps matched. Vass, is there time for a last question here, or are we-

Back to how postsecondary institutions are or aren't evolving, Paul, I think you and I have touched on this. CIGI starts to solve this gap I think with some of the explainers and that work that you take on, but there's this, I think, expectation that everyone's always learning by osmosis and staying up to date with these changing conversations and evolving policy spaces. But I do think people want to be able to sit and be like, "What is innovation policy? What is [inaudible 00:50:18]? How does it relate to AI?" And if you're not in your graduate class that you mentioned, if you're not a full-time or part-time student, it's hard to access learning spaces like that.

One other thing that we didn't really touch on that's related to exactly this and so many other things that we talked about is [inaudible 00:50:40] has an AI investment that we're making, and 2.4 billion. Most of it is on AI compute, and I've got many views on that, but I don't want to go into that. What I do want to mention is that what to me is just blatantly missing in all that is the education piece, is AI literacy. Of that 2.4 billion, there is exactly zero, as far as I can tell, that is being invested in educating everyday Canadians on what this technology is, what are its potential challenges, what are its opportunities, how can you use it? And not just everyday Canadians, but high school students and executives.

We are going to completely miss the boat as a country because without the education piece, the AI literacy piece, we're just not going to get the uptake, we're not going to get the future entrepreneurs. And that to me is the single big... And so who's going to have to do it? It'll have to be CIGI, it'll have to be universities. I'm not sure.

Interesting. We're flailing without that framework, and it doesn't appear like the politicians can deliver it either, they're scattered and really high level on it. So there's no vision to energize this. Thanks, Joel. Thanks so much for-

Thank you.

For your time for joining us.

That was really nice. That was a really fun conversation. Thank you.

I'm keen to jump in. I really liked that conversation because I thought that he was very thoughtful and balanced in his answers. We asked a lot of broad questions. There's clear there's so much we could talk about with him. We just scratched the surface on the kinds of issues that he has very interesting insights on. I thought it was interesting how innovation is one of those ambiguous concepts, which is also part of its value because it's constantly poked at and challenged. You can use it and abuse it for sure, but it's robust in its own way, even without a clear definition. So I thought that was neat.

And I thought on his model of how AI progresses from when you just start to use bigger engines to engines that are smaller scale that you can use in individual offices, and then you really reimagine things, I think is a really interesting reflection on the way technologies evolve. As long as you have that broader debate about, "Hey, what are we doing here? And is there access and is this locking in systems that we don't like and inequalities and things?" And I think he recognized that. I think on the trade thing, we didn't really get a chance to talk about it, which is a flag to me that we have to get into that in another setting because I think there's some big pieces coming within the next two years for Canada on trade. It's going to be existential for us.

Yeah. I really appreciate that he's anchored at the academy, but willing to push and pull it and recognize the forces that are reshaping universities. It's really interesting to push researchers or change the incentives towards something that is as nebulous as innovation. At the same time, it's what we want graduates to be able to do. We want people to be able to think, but sometimes the focus in a scholastic setting is purely on doing right, maybe performing more to a rubric or a template, and examples. And we don't always have templates for what we want these outcomes to be, but at the same time, it is risky and we have to be willing to invest in research that may not demonstrate an immediate return for all of these reasons. So yeah, that's something I really appreciated.

And challenging myself to see how universities are addressing labor market shifts and changing expectations and changing hopes and even who a learner is and when we learn and how, I don't think that's going to go away. If anything, I think maybe it'll be harder to recognize these institutions in a few years. At the same time, your children are going to have a pretty similar higher education experience that you did, except maybe they don't have any hillbilly notebooks or do they?

Yeah. It's been said sometimes history moves, decades are in decades and sometimes decades are in months or weeks, and it feels like we're due for one of those massive change moments, not necessarily in a good way. So a lot of this stuff matters like big time as to how things are planned out.

Policy Prompt is produced by me, Vass Bednar, and Paul Samson. Tim Lewis and Mel Wiersma are our technical producers. Background research is contributed by Reanne Cayenne. Brand design by Abhilasha Dewan. And creative direction from Som Tsoi. The original theme music is by Josh Snethlage. Sound mixing by François Goudreaut. And special thanks to creative consultant Ken Ogasawara. Please subscribe and write Policy Prompt wherever you listen to podcasts and stay tuned for future episodes.