Perspectives podcast: AI is creating ‘another industrial revolution,’ but is Canada keeping up?

Stephen Meurice: Artificial intelligence, specifically generative AI — the kind that creates content for you or answers questions based on a prompt — seems to already be changing many aspects of our lives. From how we do our jobs, to meal planning, to movie special effects, you name it.

Divya Goyal: I would say we are at the tip of the iceberg.

SM: That’s Divya Goyal, she’s an Analyst in Global Equity Research at Scotiabank. As part of her job, she writes a series called Demystifying AI. And it’s not just our personal lives that are being changed by this new technology; businesses also want to figure out how they can use it.

DG: So I think it will act like another industrial revolution where it'll truly help these companies scale up to that next level. There is a seismic shift in industries that this AI as a technology will bring.

SM: Divya is our guest today. She’ll give us the bird’s-eye-view of how Canadian businesses are approaching and adopting this revolutionary technology. She’ll tell us what sectors are ripe for disruption, what role Canada could play when it comes to homegrown innovation, some unintended consequences of this new technology and much more.

I’m Stephen Meurice and this is Perspectives.

Divya, welcome to the show. Thank you so much for coming and joining us today.

DG: Thanks for having me, Stephen. I really appreciate it.

SM: All right. So as part of your work as an analyst at Scotiabank, you've been working on a series called Demystifying AI. Are you able to give us like a snapshot of how AI is currently being used in business or industry in Canada? What's the lay of the land right now?

DG: That’s a lot to cover in one question.

SM: Sure is.

DG: But yeah, we have been actually working on this demystifying series since last year. There's a lot that has evolved in AI or artificial intelligence to put it over the last two years itself. So, AI has always been a very prominent technology. It's been around for decades, actually. For people that might not know, it's been around since the 1950s. That's AI or machine learning as a technology. But what happened was in November 2022, when OpenAI launched ChatGPT, gen AI came to the forefront. It became a very active part of everyone's life, where people could go on an app and access AI readily. Versus in the past it was almost a thing for technologists, a thing for people who are coding or software engineers kind of behind the scenes. And now everyone was using and talking about ChatGPT. So that's where AI became front and center. And you would have heard about companies — and even our Bank for that matter — talk about data analytics in the past. That was exactly a lot of work that they were doing, like, you know, AI, machine learning and using the predictive capabilities of those tools to actually bring in new solutions, enhance our internal systems. Now, when the gen AI came out, the question was how can we now bring in these LLM models which have so much more generative capabilities as compared to historical AI tools and that were kind of gen AI became front and centre. There's a lot of development that's happening on the AI and the gen AI front, right? Like pace of change is just very, very, very fast.

SM: Right? So you referred to LLMs, that's large language models, is that right?

DG: That's right, yeah.

SM: So how have things like LLMs and generative AI changed things for business or how they’re starting to look for ways they could use these systems?

DG: It's a very deep and a very important question. So broadly speaking, in my day-to-day job, I cover North American technology services companies. So, day to day, these companies, they do consulting. So, they would provide technology consulting services. They would provide the technology implementation services. Now, what happens is some of the global clients, they are now trying to understand, ‘What am I supposed to do with this gen AI technology that is out there? Like, how do we understand it? Or how do we even bring in that efficiency, enhance our productivity, improve our utilization?’ Those are all the important terms and these are very boardroom discussions. They are not simple discussions being done at the technology level only. It's not IT teams; AI and technology is out of the IT rooms. Now they are broadly being discussed across marketing departments, across the financial department, across equity research as a department to see how can we actually bring these LLM models into our system? And so, the reason I brought up technology services companies or for that matter, consulting companies. Now you think about Deloittes, you think about McKinseys of the world, they themselves are trying to understand and dive into the complexity of AI as a technology trying to unravel and unfold that technology. And then once they have an understanding, it's a big investment and a commitment at the consulting level. They then go and advise to the enterprises. And, you know, I would say it's not just about understanding technology. Once you understand the basics of technology, so how it's going to work out is, help me understand the key use cases in this technology, right? Like if I'm running a hospital network, help me understand what are the key use cases in this AI technology where I can enhance and improve my hospital system, right? How can I do better customer care? How can I serve my patient better? How can I do after patient support? Is there a cancer care treatment that I can find? There are so many questions that are hows and whys and the use cases that are currently being evaluated is where things are in a lot more that's going on.

SM: So are we just at the early stages really of businesses and industries starting to understand how they can use this technology to improve themselves or improve their processes or ultimately make more money?

DG: We are very early stages. I would say we are at the tip of the iceberg. And economics of AI is what you're trying to address here, simply put, right? So as I mentioned, when it comes to the consulting companies or when it comes to the technology consulting companies taking one step ahead. What happens is, say supposedly you are running a business, I am a tech company. I'm going to go and say, ‘Stephen, by the way, do you understand how exactly does AI work?’ Or you would come to me saying, ‘Help me understand the way how AI works.’ I’ll lay out the use cases for you. Before I laid out the use cases, it's a big commitment and investment on my front that I have already made. So I’m talking about from a revenue/cost generation standpoint. So from my standpoint there was an initial cost and so a lot of like technology companies, a lot of consulting companies are currently investing in understanding the technology themselves. Then the next step comes in as a consulting engagement where I would get engaged by Stephen's company to come and actually help them understand. What's happening today is there's a lot of proof of concepts being run across the enterprises. What those proof of concepts do is help the companies understand that are these the true use cases that I can even deploy in the long run? So today, those could be smaller engagements, more proof of concept like, but the minute those proof of concepts start to get into larger scale implementations, those small millions of dollars could convert into billions of dollars worth of engagement. The question is, are you, as a company, ready to commit to those billions of dollars? With our global economy broadly right now, in a little bit of disarray, like there's a lot happening on a global scale. Inflation's high. The interest rates are stickier than what we wanted them to be. There's a lot of hesitance in committing to these AI dollars. So there is a lot of economics in this AI when it comes to the initial cost uptake and then leading to revenue generation. But from what we'll see is different segments of the industry will take on the cost at different stages and generate revenue at different stages. But in the very long run there is a seismic shift in industries that this AI as a technology will bring.

SM: Maybe we'll just take a step back and just talk about those first couple of years really of this revolution that we're going through. Are there any examples of how businesses have already started to use these capabilities in their businesses?

DG: So, it depends upon what kind of businesses are you talking about. If you're talking about small businesses, yes, there is usage. If we're talking about start-ups, cloud native technologies that the start-ups have, then yes, there's a lot of usage of ChatGPT-like tools that are available and can be integrated into their existing systems. There can be software generation that it's being used for. So there's a lot of work that's coming out of broader generative AI as a technology. But when you start to get into some of the regulated businesses like financial institutions, some of the larger healthcare businesses and whatnot, there's a lot of data, there's a lot of privacy issues that start to come in. You think about large pharmaceutical companies for that matter, there's a lot of restrictions that start to kind of come into play with respect to can you truly use a platform that is not under your control. So, for that matter, these large enterprises are working closely with, say, Microsoft, for OpenAI and ChatGPT for that specific reason, to ensure that the technology can be now integrated within our ecosystem. Now, there are so many more steps to that integration. So one thing is integrating the technology, but then next step comes in is are data systems ready? Because some of these larger enterprises and I say that on a global scale, are still running on legacy infrastructure. And legacy infrastructure and modern technologies have a bit of a disconnect. So, where we are currently in the cycle is we need to modernize our large-scale legacy infrastructures. You'll be able to then intertwine the new AI or the LLM technology, and that's where efficiencies will truly start to come out.

SM: Right. So in order to fully take advantage of the technology, there are steps that, at least these bigger organizations, are going to have to go through in terms of modernizing their own technology before they can take full advantage. And at the smaller scale, people are already using it, businesses are using it. But again, for relatively modest purposes.

DG: Yep. That's the right way to put it.

SM: Okay. Are there certain sectors or certain types of industries where you think the potential for the implementation of AI technologies is most promising say?

DG: So simply put, any industry which is customer facing is ripe for disruption. When it comes to AI. When I say customer facing, you can think about health care systems. You can think about banking and financial services systems. You can think about retail networks for that matter, restaurants. Like anything where we are somehow as consumers interacting directly with those businesses, AI and its front-end modernization that we talked about, that's very ripe for disruption. Now, broadly, over time, these enterprises or these ecosystems will further evolve, as I mentioned, as they sort of have this new infrastructure which has moved off legacy infrastructure to a new infrastructure which is ready to get modernized, which can now be integrated. Once that happens, obviously there'll be bigger changes to come, but at the very beginning we’ll see modernization on the front end. And I think we've already started seeing like some of the cool new technology in our apps and like in our banking apps, in our health care apps, in our retail apps. Right? So there's already disruption happening. It can be a lot more multifold as AI continues to take hold. And it's important and I want to be more empathetic on this front right now here, the reason why these enterprises are not investing in the technology is not because they are not interested in investing. There is so much more stress on cost optimization, cost management just because of the global dynamics, as I mentioned, that they have to be mindful of where should the dollars be allocated. So It's not that the intent is not there, it's just the way it worked out with AI kind of came right around the time when the world's going through an upheaval. So, I think if anything, what AI would do it will act like another industrial revolution where it'll truly help these companies scale up to that next level. So, there will be investment. The pace of investment will be a little bit slower, but there will be a seismic shift in how these businesses look like in just another few years as they continue to truly allocate dollars towards AI developments.

SM: You talked about a few different sectors that are likely to be affected. First, a lot of customer facing stuff, whether it's retail or health care, things like that. Lots of people work in those sectors now. I think that's one of the main concerns that people might have as we go into this new industrial revolution, as you called it, is going to be the impact on jobs. Is AI going to steal everybody's job?

DG: So I think that's the wrong way to think about AI. I don't think AI is here to take our jobs. If anything, AI is here to enhance what's available for us, to actually help us allocate our time for something where we can truly use our intelligence to do that job. Make us more cognitively capable, is what I would say. Like when we think about AI, we talk about increased productivity, increased utilization, bringing in efficiencies. If we were to start bringing in and adapting the AI rather than resisting it, I personally believe, if anything, it'll actually make our lives and our work a lot more better and efficient. Think about it, if we had a calculator at our disposal, but we were still manually spending hours adding numbers, is that a relevant use of our time? No. Like as an equity research analyst, I can tell you there's so much time that we as individuals spend on like modeling, on writing our notes. And if we can bring in efficiencies, we should bring in efficiencies. And that applies to the retail channels, that applies to everywhere across the board. So, I think that concern is a little bit overblown as long as we are ready to upskill and reskill ourselves.

SM: Obviously AI requires tons of data, the collection of it and the storage of it. All of those things. I've seen reports on how this is causing massive new demands on electrical systems all across North America. Is that an issue with the proliferation of AI as data and data collection and storage becomes even more important, especially as we're transitioning from fossil fuels?

DG: So AI definitely increases the demand for data centres. It's directly correlated because there is a need for increased complexity in processing that data because these large language models create an increased demand for data centers. When there are more data centres, there's an increased demand for power, right? We at equity research here at Scotiabank, we've done some work on the data centres and the increased demand for power and utilities. What we also researched and come to know is it's one thing that AI will create increased demand for data centres and increase demand for power and utilities, but on the other hand, AI could also help bring in efficiencies. So, it's just with time, we'll see how things work out. But broadly speaking, we do see increased investment in data centres over the next few years. Like, you know, if it's say close to a billion today, it could be multi billion in just another few years just because of how quickly LLMs are taking hold. And every time an LLM has to look for the next generative word, it has to go back and ping its data centre. As enterprises keep adopting AI, that demand for data centres will keep going up.

SM: And where would you say Canada stands at this point, sort of in its adoption of these technologies and using them in their businesses compared to globally?

DG: So, I think Canada, it’s really no different from where the global enterprises truly are when it comes to broadly where it stands. Now Canadians can be Canadians, right? Like they could be a little bit more conservative when it comes to adopting a new technology, right? Like when I speak to, because I cover some of the North American companies and I do quite a lot of due diligence with the North American enterprises broadly beyond the companies that I cover and some of the global companies, I do see the adoption rate or even the risk taking a lot more across global enterprises. Versus I think Canadians are a little bit more like, ‘Let's try to test the waters and see how things are kind of shaping up and then we will eventually get there.’ Which is not a bad approach, to be honest in the short-term, because AI, the pace at which this technology is changing and evolving, there's a lot that can change from where things are today to where things could be in six to eight months. Now, having said that, there is a reactive and a proactive approach. You have to be somewhere in the middle. In a technology like AI, which is changing at the pace it is, being proactive is important, but you cannot just wait to be reactive as well. So it's important that enterprises continue to keep testing the water, see what's out there, proof of concepts, like maybe slowly start committing and investing in some of these technologies so that they are truly not left behind.

SM: And where does Canada stand when it comes to being involved actually in the development of these new technologies? It kind of seemed to me that there was talk few years ago, Montreal seemed like it was becoming something of an AI hub. I'm not sure whether that's still the case anymore. Toronto certainly has some activity as well, but mostly what you hear about is still Silicon Valley. How much work is going on in Canada to contribute to the development of this technology?

DG: So when it comes to Canada, I think Montreal still definitely is a hub when it comes to the development of the AI. We recently at Scotia hosted our private tech conference in Montreal. We had some really talented companies come and present to us, the founding fathers of the Deep Learning Movement, Professor Yoshua Bengio. He's a professor at University of Montreal. And other than that, there's a lot of other work that's being done in Montreal when it comes to AI research and AI development. CGI, one of the companies I cover, they are based out of Montreal. They have committed a billion dollars to AI-led work and development, so an internal investment of a billion dollars in AI is what CGI has committed to. So, there's a lot of interesting work that's being done in Montreal. There's a lot of new companies that are coming out. Toronto is not far behind at all. Creative Destruction Lab is doing a lot of work out of University of Toronto. I have a lot of respect for Professor Ajay Agrawal. I happened to be one of his students way back when, so they have done a great job. Vector Institute in Toronto is doing a lot of work. Recently there was a company, Sanctuary AI that I was reading about. They partnered with Accenture and they are the ones that are developing humanoid robots and stuff. So there's a lot of interesting work. Cohere, I shouldn't forget to mention, they are one of the leading LLM companies on a global scale, so there's a lot of interesting technologies that are getting developed in Canada. Now, when you ask me about development, I think there's a lot of innovation and development that's happening in Canada. What concerns me is just the scalability. You know, is it scalable in Canada? Once they reach a certain level, would they continue to stay within Canada or would they have to look outwards? And that's when I think Silicon Valley just starts to look more appealing and more attractive from an investment standpoint.  When these companies are seeking dollars, do we in Canada have enough investment dollars? And I think that's where the disconnect sort of slightly comes in, where we end up losing some of our talent to our neighbours to the south.

SM: So, we've talked quite a bit about the early stages of this revolution, what we're going through now, might see in the next couple of years. Can you tell us what's this going to look like in ten years? How is this going to reshape business? How is this going to reshape regular people's lives?

DG: I think the world's going to be really different in ten years. That's just me, the way I see things shaping up. The pace of AI, it’s something that we have not seen in the past, ever. So, there's a lot of curiosity on AI, there's a lot of concerns around AI and there's a bit of panic around AI. I think what is important is that we as individuals, our companies and governments need to understand the implications of AI. How does it impact our companies and industries? And where there is a potential for development, we need to continue to invest in adopting this technology rather than being fearful of this technology and embrace it. Upskill ourselves and make sure we are a part of the ecosystem and not looking at it as an outsider, but being a part of this system. So, I think in ten years we'll be in a new world. People can actually do more things a lot more efficiently. I think it’s going to be for the better.

SM: Right. And from a business perspective, I mean, your job is to assess companies and what their future looks like. You're an optimist about what the impact of AI will be on business and industry?

DG: So, yes, I am an optimist when it comes to AI. I do think AI though, will need a lot of investment and that investment will vary based on the size of company you are. So, there's a lot of work that will go into getting AI to the levels we are thinking, but people just have to keep their minds open. They just have to continue to keep investing in the technology. We as individuals and the governments of the world, global enterprises, they all have to take this as a new change and a positive change. And once we start looking at it that way, once we start truly analyzing the use cases that can make better companies, better economies and ourselves as better individuals, I think AI will just work in our favour.

SM: So, ChatGPT, obviously it changed everything for everybody. It was the cool new technology. What's the next thing coming out? What are you keeping your eye on?

DG: Interesting question. So, in AI and machine learning, what currently happens is there is a term called reinforcement learning from human feedback. So once the ChatGPT or say an LLM produces a certain response, there is a continuous training that's happening to make sure that the LLM is not hallucinating. There are no biases. So, there is a lot of internal technological development that's happening. I think in future, deep unsupervised learning is what we could potentially start to see these models going towards, which will use some of the deep learning principles to learn in an unsupervised way. Meaning without human guidance or labels on the data that is fed into AI systems, and that is then being used to train them on the horizon.

SM: So, they'll learn on their own?

DG: They’ll learn on their own.

SM: Divya, I want to thank you very much for coming. I really appreciate it. It was a really interesting conversation.

DG: Thanks a lot, Stephen.

SM: I've been speaking with Divya Goyal, Analyst, Technology – Software & Services in Equity Research at Scotiabank.