The Empire Club of Canada Presents
Charting Canada’s AI Future: How to Build a Resilient Framework
Chairman: Sal Rabbani, President, Board of Directors, Empire Club of Canada
Moderator
Jordan Jacobs, CEO, Managing Partner, Co-Founder, Radical Ventures
Distinguished Guest Speakers
Angus Lockheart, Senior Policy Advisor, Dais at TMU
Nicole Foster, Director, AWS
Panellists
Tony Gaffney, CEO, Vector AI Institute
Chris Walker, CEO, Untether AI
Mara Lederman, Co-Founder, Chief Operating Officer, Signal 1 AI
Martin Kon, President & COO, Cohere
Head Table Guests
Dinar Ahmed, Partner, BDC Seed Venture Fund
Carrie Bois, CEO, C+B Consulting, Treasurer, Empire Club of Canada
Lauren Bull, Transaction Services, KPMG LLP
Jimmy Di, Master of Mathematics in Computer Science Candidate, University of Waterloo
Isaac Olowolafe, Co-Founder and General Partner, BKR Capital
Introduction
It is a great honour for me to be here at the Empire Club of Canada today, which is arguably the most famous and historically relevant speaker’s podium to have ever existed in Canada. It has offered its podium to such international luminaries as Winston Churchill, Ronald Reagan, Audrey Hepburn, the Dalai Lama, Indira Gandhi, and closer to home, from Pierre Trudeau to Justin Trudeau; literally generations of our great nation’s leaders, alongside with those of the world’s top international diplomats, heads of state, and business and thought leaders.
It is a real honour and distinct privilege to be invited to speak to the Empire Club of Canada, which has been welcoming international diplomats, leaders in business, and in science, and in politics. When they stand at that podium, they speak not only to the entire country, but they can speak to the entire world.
Welcome Address by Sal Rabbani, President, Board of Directors, Empire Club of Canada
Wow, good afternoon, and welcome to the Empire Club of Canada. My name is Sal Rabbani, and it’s an honour to stand before our community, both in person and virtually, as Chair of the Board of Directors of the Empire Club of Canada.
To formally begin this afternoon, I want to acknowledge that we are gathering today on the traditional and treaty lands of the Mississaugas of the Credit, and the homelands of the Anishinaabeg, the Haudenosaunee, and the Wendat Peoples. We encourage everyone to learn more about the Traditional Territory on which you work and live.
Today, we bring you an event on one of the most strategic and critically important topics for our future: artificial intelligence. Our leading panellists comprise top AI startups, researchers, investors, and we’ll dive into key priorities as Canada looks towards the future of AI. There’s a lot to consider, from government investment, to the state of adoption, to the unimaginable way it will impact our economy, our workplace, and our lives.
The Empire Club is a not-for-profit organization, and we’d like to recognize our sponsors, who generously support the club, and make these events possible and complimentary for our online viewers to attend. Thank you to our Lead Event Sponsors, Amazon Web Services AWS; and thank you to our Supporting Sponsor, Osler, Hoskin & Harcourt LLP; thank you to our media partner, The Logic; and lastly, thank you to our Season Sponsors: Amazon Web Services AWS, Bruce Power, and Hydro One. As always, we accept questions from the audience for our speakers, and you can undertake to scan that QR code found on your program booklet.
Canada stands as a global leader in artificial intelligence, driven by renowned research institutions, robust government support, a thriving ecosystem of startups, and its ability to attract top talent worldwide. With a commitment to ethical AI development, and active participation in international collaborations, Canada’s leadership in AI research, industry collaboration, and talent attraction is undeniable. The last two statements were written by ChatGPT. And my prompt was a simple question: Is Canada a leader in the AI space? When I asked for just one paragraph—otherwise, I would have kept you here for half an hour, you know, reciting all of the advantages that Canada has on the global AI stage—you know, besides the fun of using a conversational AI model, my point is that we’ve been talking about AI for years now, but it’s only recently that we’re starting to see it in action, and kind of grasp its applications and benefits.
My second point is that everything that ChatGPT generated as an answer, in this instance, I believe to be true. Canada does have the research talent and startups to become a leading AI supplier. But that’s not enough, if we truly want to lead. And artificial intelligence is one area where we can absolutely lead. So, what’s missing? How can we create ideal conditions, and partnerships, for Canadian AI innovation to thrive, access capital, and become mainstream? From a larger business and organizational perspective, is our data ready for AI? Are our people ready for AI? And how can we prepare for the rapid changes that lie ahead of us?
I hope today’s panel will help us find some answers to these questions. But I want to stress how important education, literacy, and good old change management are; how our ability to learn, adapt, and be agile, can help us make the most of what AI has to offer. Personally, I’m also hoping to hear from our distinguished panellists where they stand on the recently announced Voluntary Code of Conduct around how advanced generative artificial intelligence is used, and developed in this country. Some are concerned that this code could be limiting innovation, while on the other side, we hear a community voicing concerns regarding ethics, copyright, transparency, bias. And if there’s anyone in the audience who thinks that this is a topic that doesn’t concern you, this is just not true. It concerns all of us. Many of the promises that AI was making a few years ago are realities today.
I continue to be amazed by how AI can go from automating small tasks to huge transformational impacts. From writing that small paragraph for my remarks, to analyzing stock market data for trading insights, and to drive decision-making in health and improving patient care, artificial intelligence can improve our lives. It can boost productivity, it can unleash creativity, it can open doors in science, business, social things, diversity, inclusion. The potential that we’re looking at is just incredible. This is why we’re at a crossroads, and this is why today’s conversation is so important.
AI is supposed to be a leading driver of our times, and we can build a world-class AI ecosystem in Canada. We can lead globally in this space, and also use AI as an opportunity to share our Canadian values with the world. We also have a responsibility to build trust, and make sure AI is deployed ethically and responsibly. And so, before I invite our first speaker to the podium, I’d like to take a moment and acknowledge the bright young minds who are joining us today from the University of Toronto, Humber College, and Toronto Metropolitan University. Thank you for being here. This is one of my commitments. And then we were talking about it at our table earlier, just, you know, engagement from the next generation is really important, and I appreciate you all taking the time.
It’s now my honour to invite to the stage Angus Lockheart, who’s the senior policy advisor at the Dais at TMU, to explore the AI conversation further. Angus, welcome.
Opening Remark by Angus Lockheart, Senior Policy Advisor, Dais at TMU
Thanks to the Empire Club for having me. As Sal said, I work at the Dais, a think tank at Toronto Metropolitan University. One of our focuses is understanding the impacts of emerging technologies like AI in Canada. The research I’m going to be talking about today wouldn’t have been possible without the support of Amazon Web Services, [indiscernible], and the Québec Innovation Council. So, I’d just like to acknowledge their support, and continued support, of the Dais upfront.
So, when we talk about the commercialization of artificial intelligence at the Dais, why do we care? Well, we care because—and this is a problem Canada’s been facing for decades—there’s declining productivity growth. Back in 2022, Chrystia Freeland acknowledged that we’re falling behind, when it comes to economic productivity, and called it both a well-known Canadian problem, and an insidious one. And just for reference, this sort of started probably around 2000, when we started diverging from the United States—and we ended up our growth is now sort of half the rate of that in the United States. So, this is an ongoing problem for the last two decades. To us, artificial intelligence represents a path to challenging this. It’s an opportunity to fight labour shortages, while increasing productivity and bolstering Canada’s competitiveness. But that’s only assuming that we can adopt artificial, artificial intelligence successfully in businesses—and that’s a big assumption.
So, what does Canada look like right now? Well, our research relies on data from Statistics Canada, and in particular, the survey of digital technology and internet use. And the number we came to was, 3.7 percent of businesses in Canada are already using artificial intelligence. That’s an improvement from 2019, when only 2.4 percent of businesses were using it, but it’s not dramatic. When it comes to who’s adopting artificial intelligence, the biggest differentiator is the size of a business. Adoption in Canada is really concentrated in large businesses with more than 100 employees, leaving small and medium businesses behind. In large businesses, nearly 20 percent have adopted artificial intelligence already, compared to less than 3 percent among small businesses. The silver lining here is that, when it comes to large businesses, they actually employ a larger percentage of Canada’s workforce, so around 15 percent of workers in Canada are exposed to AI in the workplace. But it’s still a long way to go. And we can only rely so far on large businesses getting us there.
And the challenges facing small businesses in Canada are twofold. First off, small businesses don’t know what they don’t know. The cost of getting informed about AI systems and figuring out what they could do for you is high. For most businesses in Canada, the question is still, “How can AI help my business?” not “How do I implement it?” Our research suggests that more than 70 percent of businesses are still trying to figure this part out. However, once they’ve identified an AI system or tool that can work for them, [indiscernible]’s still face significant challenges developing their data infrastructure, and their in-house technical skills. The cost of entry on both doesn’t scale linearly, making it more affordable for large businesses to get started. Our research suggests that when businesses actually do continuously invest in upskilling, they’re significantly more likely to adopt AI.
So, zooming out, how does Canada compare globally? Well, the great thing about the data source we use is that it’s developed in part by the OECD, which means we have a clean comparison across countries. When you bring it together all the 38 countries that use the same framework, Canada comes in 20th, and the gap between us and the leaders is massive. Our data shows that Denmark ranks first, with a lot of other European nations near the top. And when you talk about the actual adoption rates, the difference is stark. Businesses in Denmark are nearly five times more likely to have adopted AI today than Canada, and other European nations are still three or four times more likely.
So, it’s a real challenge for Canada. But I think there’s some room for optimism. Canada does have a great history with artificial intelligence; it’s just mostly been on the research and development side. We were early actors launching the Pan-Canadian AI Strategy before most countries developed their own. But given the state of AI globally, the focus needs to shift towards commercialization. On this, we at the Dais actually do have some work coming out soon, on AI compute capacity in Canada, suggesting ways for Canada to become more competitive globally.
So, to sum up the challenges that face Canada, to stay competitive globally and improve the quality of life here in Canada, we need to support the adoption of AI in businesses of all sizes, including both building demand for AI, and helping bridge the skills and data gap that exists for small businesses. We also need to do it in a framework of responsible AI. An Ipsos survey found that only 32 percent of Canadians believe that there are more benefits than drawbacks to AI, putting us second-last in the 28 countries listed. We can’t forge ahead with commercialization, without protecting against the real risks AI poses, and getting Canadians onside.
So, without further ado, I’d like to welcome Jordan Jacobs, the moderator of today’s panel, to the stage. Jordan Jacobs is a managing partner and co-founder of Radical Ventures, a VC firm that invests in AI companies, with offices in London, Toronto, and Palo Alto.
Jordan Jacobs, CEO, Managing Partner, Co-Founder, Radical Ventures
I think we’ll do this this way. I’m just going to invite up all the panellists at the same time so that we can maximize our time together. So, Mara, Martin, Tony, and Chris, why don’t you all come up? So, grab a seat wherever you like.
Chris Walker, CEO, Untether AI
Musical chairs.
Jordan Jacobs
Okay, I’m gonna ask you to each just introduce yourselves very quickly, and then we’ll kind of dive into the conversation. So, Tony?
Tony Gaffney, CEO, Vector AI Institute
Sure. Tony Gaffney. I’m CEO of the Vector AI Institute, here in Toronto.
Chris Walker
Chris Walker, CEO of Untether AI, headquartered here in Toronto, and we make chips.
Mara Lederman, Co-Founder, Chief Operating Officer, Signal 1 AI
Mara Lederman, co-founder and COO of Signal One AI. And before that, spent 19 years as a faculty member at the University of Toronto.
Martin Kon, President & COO, Cohere
Martin Kon, President and COO of Cohere. We have offices in Toronto, as well as San Francisco, London, and I believe in the next couple of days, New York City. Recently returned to Canada after most of the last 30 years in Europe and the US, most recently as the CFO of YouTube.
Jordan Jacobs
Thank you. Okay. So, let’s just do a little primer on what AI is, because we hear a lot about it—maybe not as much as we hear about Taylor Swift, but it’s close. The way I think about AI is it’s, essentially, modern AI is prediction software, and it’s the ability to take in vast sums of data and make better predictions out of it than we could previously with other systems, or even humans can do. And that can be with language, which we’ve seen with Cohere, it can be with images, it can be with a combination of different types of data. So, in that context, what’s happened in the last 14-15 months since ChatGPT, I think, is the world has woken up to this potential. I would say that we all have previously had the experience of trying to explain to people why AI is real and what it means, and now, everyone can go play with it, and that’s kind of changed everything. It’s accelerated demand for products, I think, dramatically. I mean, certainly, we’ve seen it’s given a lot more attention to the research community, as well. Before ChatGPT, we kind of all existed in a world where we were, I think, having to, you know, say to people two things: one, “AI is real and it’s impactful”—and people kind of believed it or not—and then, you know, for those of us who were raising money, we were raising money and saying you have to, “You should give it to us, we’re going to make a difference.” I think the first part of that we don’t have to do anymore. Everyone now believes it. And it’s from the boardroom to, you know, the boardroom table, where the discussion is around AI, back to the, you know, dining room table at home, where the CEO or chairperson’s kid has just done a dinosaur essay with ChatGPT for fourth-grade science.
So, in that context, kind of everything has changed; discussion’s changed dramatically. There’s opportunity, I think, everywhere. The technology is essentially like electricity, in that it’s going to change every industry; it’s going to touch everything. You all are in very different industries. So, maybe we can talk about the impact through the context of your individual organizations. So, Martin, why don’t we start with you? Can you talk through both what the AI is that you are producing, and where it’s going, how it’s being used, and how it differentiates from what your customers could do before?
Martin Kon
Sure. First of all, an honour to be here—so, I’m really excited, a bit nervous with all these people looking at me. But Cohere is a foundational Enterprise AI company. So, we, we build proprietary foundational large language models that enable these incredible products to be built. We do not make apps ourselves; we do not do consumer; we don’t have a consumer chatbot, which is why our brand might not be as well known out on the street as some others. But what we do is build the models, the capabilities, the tooling, the inference frameworks, fine-tuning, et cetera, to enable enterprises to create incredible products, and create real business value at massive scale. That means they need to be efficient. I sometimes use the analogy of going for a test drive in Bugatti and being impressed by the top speed, and then realizing it would be a bit expensive to equip your salesforce with a fleet of Bugattis—also, they only have, you know, two seats and no trunk. We really think about what companies need, what enterprises need, to deploy at massive scale, so that hundreds, thousands, tens of thousands of employees, or millions of customers, can use these capabilities. And that’s both the generative models that I think many of us know, through things like ChatGPT, you know, “Write me a poem about my parakeet.” Great. But also, the retrieval models—or the representation models, sorry—which help with search and retrieval. And that is a massive unlock that is not talked about as much. But I can tell you, enterprises are even more excited about that capability of being able to search very quickly through enormous masses of very sensitive internal data that’s, you know, spread all over the place in silos, retrieve that, and be able to use that information to make better decisions, and so on.
So, I got to talk a little bit—maybe as a follow-up question when we get into it—but how transformational this is for enterprises and the entire world. But I’ll let my colleagues continue.
Jordan Jacobs
Just, why don’t we continue with that for one second. How big is the opportunity? Like, what is the market opportunity, here?
Martin Kon
Yeah, so, as you can maybe tell by the colour of my beard, I’ve been around the block a few times. I started my career in 1993, when I graduated from McGill. The Mosaic Browser came out in 1993. This is the biggest transformation disruption to everything that the world’s doing since the Mosaic Browser came out in 1993. That was when the internet sort of started to become mainstream, and nothing was ever the same again. We can’t even imagine a world without the internet. But it existed. I didn’t have email when I was at McGill. And if you look at how much the world changed, let’s say from ’95 to 2005, every single enterprise just fundamentally changed everything they did. And those that didn’t, don’t exist anymore. Those that did it well are extremely successful. Does anyone remember Blockbuster? Doesn’t exist. Anyone know what….
Mara Lederman
But I remember them.
Martin Kon
You remember them. Anyone, you know, use Netflix? There you go. So, someone on, you know, the other side of that. Now, there are other companies—Microsoft, they existed before. But they pivoted, and now they’re worth three trillion dollars. They’re the most valuable company in the world, more than Apple. And so, the same will happen now, in terms of enterprises that embrace this, and are quick, and innovate, versus those that don’t. I think just to put a, put a sort of a human point around it, this is just the way that humans will demand to interact with computers, period. The same way that we, you know, demanded to use these things, and touch screens, and so on, and the same way that we, you know, used a browser, and a mouse, and clicked through Internet Explorer back in the day, and Chrome now, mostly. This is the way that consumers will talk to their service providers, this is the way that employees will talk to their enterprises. It’s just, it’s already happening. And it is the biggest disruption, transformation, and opportunity for job creation and value creation in 30 years.
Jordan Jacobs
Okay. So, we’re going to come back to some of that and then pack it a bit. Mara, you’re in a completely different space.
Mara Lederman
Yep.
Jordan Jacobs
Tell us what you guys do.
Mara Lederman
So, we build technology that allows health systems to integrate AI into clinical care. And I think it’s probably uncontroversial to say three things about AI in healthcare. Healthcare probably is the biggest opportunity for AI—all types of AI, not just what we’re doing. Two, almost no industry is as behind as healthcare in adopting AI. And three, we all believe we’ll get there, but kind of none of us know exactly how. And so, it’s, it’s one of those businesses where, you know, you can say we’re building AI for hospitals, like, “Oh yeah, hospitals are eventually going to have AI, but I don’t know how they’re going to get there.” And that’s the problem we’re trying to solve. We’re trying to build the technology system, both the actual AI applications, but the underlying platform that manages those applications, that monitors those applications, that supports AI governance.
If we sort of think about how big—you know, you asked, you asked how big is the opportunity—I think the opportunity in healthcare, it’s not just that it’s massive; it’s that it’s entirely essential. And you can look at what’s going on in healthcare systems around the world—and maybe, Canada really is the worst example—but at least here, we are in total crisis in our healthcare system. And I think you open the newspaper, you—well, I guess you read the newspaper, you don’t open it—you watch the news, you step into an emergency department, and it is smack right in your face, the fundamental problems we’re having delivering healthcare in Canada.
We have a huge mismatch of supply and demand. Jordan knows I’m an economist, I always talk about things where I can in economic terms. We do not have the ability as a country with our physical infrastructure and healthcare and our workforce, to deliver healthcare, to meet the demand for healthcare in our population. We are not going to build our way out of that with more hospitals, we are not going to hire our way out of that, and we’re certainly not going to get there by just, like, shuffling nurses from one place to another, you know, and changing what we pay them for different types of jobs. So, the only answer is technology. Not just AI, but AI in particular, is incredibly well-suited for expanding our capacity to deliver healthcare to a growing number of people. We need to bring supply-and-demand, basically, back into match, and now, start thinking about all the things AI can do. You know, AI can be used to determine who walks into an emergency department and doesn’t need to be there, and they can be funnelled to a different place of getting care. Take it a step back, you can have a digital application that does that before people even show up. AI can be used to shorten hospital stays—that’s what we’re trying to do on one of our applications, to get people out of the hospital sooner—AI can be used to screen people who can finish their care episode at home. We have a few hospitals in this country launching what we call “Hospital at Home Program.” Fantastic. People go home sooner, their bed is opened up for someone who’s sicker, the cost of that episode goes down by 30 percent, and AI helps in two ways: it finds the people who are suitable, and then it monitors at home. So, if something goes wrong, their care team gets alerted, and they come back. Over and over and over again, you can think about how AI is going to transform the delivery of healthcare. And I think once you start thinking about it that way, I don’t think you can imagine any other way out of this problem, other than one that is entirely enabled by technology.
Jordan Jacobs
And so, there’s two points there. One is, it makes the delivery of this much cheaper, and….
Mara Lederman
And different. It’s cheaper because you can deliver it in different ways…
Jordan Jacobs
Right. In better ways, which result in….
Mara Lederman
…in lower cost ways.
Jordan Jacobs
But ultimately in more health, longer lives. So, you’re achieving, in a single-payer system, the two things that we should all want, which is a more efficient system that actually delivers better healthcare.
Mara Lederman
Exactly.
Jordan Jacobs
Okay. And you’re deployed in multiple hospitals already. So, let’s come back to…
Mara Lederman
Yep.
Jordan Jacobs
… some of that, and some of the experience that you’ve seen. Chris, again, doing something completely different…
Chris Walker
Right.
Jordan Jacobs
…but also AI-focused. Tell us what you’re doing.
Chris Walker
For us, at Untether AI, we’re the Silicon. So, we’re what the AI models run on, how it gets deployed—so, how we accelerate that and, and make it real. And I’m glad you talked about AI in terms of, you know, multiple different facets, including industry and vision. So, one way that we help AI become real in the world is things like taking a vision, being able to process it faster, so a street corner can be safer. Agriculture technology, you don’t think about AI in that, but if you can have an autonomous tractor that instead of using pesticides is using laser to zap weeds, AI is necessary to do that at speed, and at a pace to make it productive. And then, I think, for, what’s also very transformative is in the generative and language model space. It has the opportunity to take small businesses, and let them compete like big businesses. And this is something as important as when the internet opened e-commerce, you know, opened the world to people, through the internet. The generative tools, the things people are going to do to build, and use code to build up their applications, it’s going to make them appear and compete on the world at a much larger scale, as a, as a startup.
But what’s important, as all that comes together, as we look at this mass, mass deployment of AI that can all benefit us, there’s real implications to it, in terms of what’s the power required, what’s the heat required. And this is something that we all run today on, basically, a core architecture and way of processing that’s decades old. And our approach and what we need to do is find—and we deliver solutions to do that much more energy-efficiently, while preserving accuracy and speed. So, that 20-, 30-query ChatGPT that you do, you don’t realize that the data centre behind it may have burned a half a litre of water to fulfill that. And so, we think, as it impacts all of our lives to realize this potential. We also have to address the underlying capabilities that these things run on, and make sure it’s effective and efficient, as well.
Jordan Jacobs
Just to go a little bit deeper there, I mean, we’ve seen one of the first true beneficiaries of this wave of AI has been Nvidia and its stock price, right?
Chris Walker
Sure.
Jordan Jacobs
The company’s worth 1.7 trillion. Demand has exploded, because people are, their customers are basically buying the chips, in order to train these systems. The even bigger opportunity is, once the systems are in place, it’s what you’re talking about in production. It’s the demand for all of that compute, after the training is done. I mean, the training will continue inevitably. But once every one of your customers’ customers is using it, and once every one of your customers is using it, I mean, there’s this explosion of demand. So, the purpose of what you guys are focused on is delivering chips that are much more power-efficient; can operate, as you say, on the tractor, or on the edge, without having to be in a data centre and connected to some power source. So, you can actually run these very large language models, or decision systems, or computer vision systems, in a way that makes sense. We should talk more about chips, because they are kind of the chokepoint of the entire global economy at the moment, geopolitically and economically. So, we’ll come back to it. But just tell us for a second, where are your chips manufactured?
Chris Walker
So, we design hardware and software here in Toronto. We manufacture in The Foundry Network, TSMC—like 90-plus percent of everyone who’s like us. And that’s a critical part is, you know, if there’s a geopolitical about it, there’s different foundry networks. We obviously look at things like that. But most companies are [indiscernible], and you know, all rely on one centre point of production across the world.
Jordan Jacobs
Right. Just to be clear, over 90 percent of the leading AI chips in the world, from Nvidia, Google, everybody else who’s doing advanced chips are manufactured in one facility in Taiwan.
Chris Walker
Correct.
Jordan Jacobs
That is a subject of the biggest geopolitical tension between superpowers, currently, in the world. So, an interesting flashpoint, there’s been lots of prediction that the next World War will be fought over TSMC. And so, we’ll come back to that. The US is trying to mitigate that by enacting the Chips Act, and building foundries in the US. But the capacity to build chips, even when, once all those factories are up and running, will be 10 percent of what TSMC does now. So, all the advanced AI in the world is dependent on these chips, the chips are all manufactured in one place. And so, it’s an issue we should come back to. Because it has created a real competition between countries, which we’ve been talking about a lot lately. Tony, before we get more into that—you’ve been involved in that discussion—tell us about what you do at the Vector Institute.
Tony Gaffney
So, I think the best way to start talking about Vector is to call out what Jensen Huang, talking about Nvidia, called out at his recent visit, which is it’s really important to understand that Canada is globally the epicentre of the research and development of AI. And out of that work, Vector was born. So, thanks to Jordan and another founders, Vector was founded to ensure that we continued that quality and pace, and leadership, from a research perspective, with a mandate to build talent here. And it’s important, also, to understand that we are the fastest-growing AI hub over the past five years in North America, and one of the few in the world, with the net inflow of AI talent. That gives us a terrific platform to work off.
And our focus is harvesting that advanced frontier of research and adoption with our community, to drive economic benefits, and also societal benefits, to the benefit of Canada as a whole. And a huge enabler to make that happen was the foresight of [indiscernible], back in the early days, to support the implementation of a national strategy in Canada, and that led to the building and evolution of our three institutes, ourselves, Vector, Mila, and [indiscernible].
However, despite being one of the first countries in the world to have a national AI strategy, we all know that if you stand still, you start slipping. We’re slipping. So, there’s a need to refresh our national strategy. And you know, two particular areas already called out, Angus, in your commentary, is we need more infrastructure, compute in particular, and we need better access to data, including in particular sovereign data, such as climate data, health data, cities data, and so on, to help advance the adoption of AI here, locally. So, Jordan?
Jordan Jacobs
So, the things that we believe you need to be successful in building an AI company are the best people…
Tony Gaffney, CEO, Vector AI Institute
M’hm (affirmative).
Jordan Jacobs
…and that’s where Vector comes in, and the other institutes that were created by the Pan-Canadian AI Strategy. How many master students a year in AI are you graduating?
Tony Gaffney
So, our target when we set out was to graduate a thousand work-ready Masters in AI. And we’ve now exceeded that, and have a significant community of, of graduates. And what I love and get very excited about is that the percentage of those who graduate and get work here in Canada is in the high 90’s. The number who have left, or who are still here after several years, is in the 90’s as well, which is really encouraging, and reinforces the fact that we have a net intake. But as I said, we can’t take that for granted and stand still. We’ve got to modernize our overall approach and strategy.
Jordan Jacobs
So, so people is number one. And I think that Vector and Mila in Montréal, and [indiscernible] in Edmonton, and other organizations, have done an excellent job of reversing the brain drain…
Tony Gaffney
M’hm (affirmative).
Jordan Jacobs
…and helping to create the supply of talent that is necessary to draw companies in.
And we’ve seen an explosion of companies in Toronto, in particular. I think there’s about 50 big company labs. Whereas when we were running our company, Layer 6, there were none. And there are hundreds and hundreds of AI startups. Back then, I think we were the only one. Our motivation for creating Vector was when we were recruiting amazing people from DeepMind in London, or from Google Brain in Mountain View, what they would say is, “I’m willing to quit my job, and so is my spouse, and we’re going to pull the kids from school, and sell our house, and pay an exit tax, and move to Toronto. But if it doesn’t work out with you, there’s no other jobs. So, I can’t take that risk. I can’t come all the way and make all those sacrifices, and then if it doesn’t work out, there’s just nothing else to do.” I think that’s been resolved by having a centre of gravity that produces talent like Vector here, that has aggregated all these big companies, and small companies, and medium-sized companies around it. It was one of the decision points when we were talking about creating Cohere, picking between San Francisco, London, and Toronto, was that access. So, that, I think, has been—I don’t want to say completely solved, but we are doing as well as anywhere in the world in producing extraordinary talent.
And that goes back to the starting point of Jeff Hinton, locally. For those who haven’t heard that name, Jeff is a professor at the University of Toronto. He’s responsible for a lot of the breakthroughs that have led to modern AI. His students—he won a big competition in 2012 with two of his students, called ImageNet, run out of Stanford. That resulted in his students being bought up and becoming the heads of AI at Apple, Facebook, Uber, Tesla, OpenAI, leadership at DeepMind, Google Brain, Microsoft Research, and [indiscernible]. Like, all the top AI labs, at the top tech companies in the world, were led by people from one lab at the University of Toronto, you know, a kilometre-and-a-half away from here—in a very Soviet-looking building. Very few windows. But they really went on to lead the world. And I think our story is very Canadian. But people outside of Canada know it better than we do.
Tony Gaffney
Yep. Yep.
Jordan Jacobs
In Silicon Valley, everyone knows that story. Even in China, people know that story. Here, we don’t know it as well. Why is it—and when you’re experiencing this selling around the world or domestically—why do you think it is that Canadians are not as aware of this, or what the potential of AI is? I’m making an assumption, you can tell me if I’m wrong. Why are other organizations and other places willing to adopt this faster? And tell us if that’s true or not in your own cases, individually. Do you want to start, Martin?
Martin Kon
Sure.
Jordan Jacobs
He comes from Google—so you can use your own personal experience.
Martin Kon
Yeah, no, I think, first of all, I really enjoyed some of the things that Angus said. I can’t remember the exact number, so I’m looking forward to reading the report, but you know, Canada, compared to a place like Denmark. I think Canada, just chronically, is a late adopter of technology across the board, Canadian companies. I think it is something that, you know, I think we’re very happy to—and I’m very proud Canadian; I am now American as well—I think we’re very happy to say things like, “Oh, you know, The Rock is part Canadian and Justin Bieber’s Canadian,” and all these kinds of things. They’ve all, you know, done, made it big in the US. I was in the US for many years—Aiden Gomez was the co-inventor of the Transformer, our CEO in Mountain View. And I think one of the things is, maybe there’s this, there seems to be this chronic inferiority complex in Canada—I’m a Canadian, I sort of feel that as well—that everything in the US is better. And the, we feel that now. So, 99.9 percent of our revenue comes from outside Canada. Not because we don’t want to sell to Canadian companies, but because Canadian companies don’t even call us. It’s changing now, because they see that Larry Ellison at Oracle has bet on us, Salesforce has bet on us, SAP has bet on us, you know, a massive investment bank in New York that doesn’t let me say their name out loud ,has bet on us. And then, you know, Deutsche Telekom’s VC arm invested in us. And then, suddenly, people here take notice, like, “Oh, wow, you know, let’s talk.” You know, we didn’t have any calls from the Canadian telcos, even though we personally know the CEOs, but we had SK Telecom, and Vodafone, and Verizon—and I was at a dinner with the AT&T board, we were with Tim Höttges at Deutsche Telekom when he visited the Bay Area. And I think in Canada, we tend to look south of the border, everything’s better there.
I’ll tell you a very true story—which is amusing but also a bit sad. Our headquarters is in Toronto, 40 percent of our staff are here, 40 percent in the US, and 20 percent across Europe. A lot of our leadership is up and down the West Coast of the US. But a lot here. We now have in our PR, our press releases, Toronto and San Francisco as our dual headquarters. That’s not for Americans; they do not care. It’s for Canadians. Because that way, Canadian companies take us more seriously, if they think that we’re dual-headquartered in Toronto and San Francisco. That is so sad. It’s the only country in the world that does that. I can tell you that major German enterprises are all using Aleph Alpha. I don’t think they have anywhere near the same quality technology, but they’re a local hero, and they’re using them to support the local folks. French companies are using [indiscernible]. The governments are supporting both of those very deeply, in a way that, as of up to now, the government here hasn’t. And so, I think there’s part of this inferiority complex, and that’s why people here don’t know that story. Because they assume, well, obviously things are better in, you know, at Meta. It’s like, yeah, but it’s run by a Canadian, you know, Yann LeCun—he happens to do it out of New York, Ilya at OpenAI happens to do that in the Bay Area. So, our co-founders came here and said we want to build a generational company in Toronto with—I met some of the UofT folks, give them a shout out in the back there—to be around people like that, because they want to give all these incredible this incredible talent the opportunity to stay here and build a global company like Shopify, that just happens to be headquartered here.
But I, I so I think it’s two things. One is, generally Canadian companies adopt tech later, and that is going to be absolutely deadly in this time. I mean, whether it’s healthcare, whether it’s the productivity crisis that we have in this, you know, in this country. We’re not going to solve that through immigration alone. We need to do that through using AI, to have a step change of productivity. If companies here are late to adopt, maybe if they’re protected by government regulations in certain sectors—another thing that gets a bee in my bonnet—but barring that, we’re just not going to compete. And then the second thing is, where do we look to get that technology? There are some incredible world-leading companies that are literally down the street. There’s another example where there was a big event—I won’t say who was hosting it, but they had all these known Canadian leaders. And they had Sam Altman on screen, from San Francisco or wherever he was. And Aiden Gomez was sitting a block-and-a-half away. Sam didn’t invent the Transformer—he just is in the headlines a lot—Aiden did. Why, you know, some of those things. Like, why would you not want to work with the team, the Chief Product Officer who came back to Canada after 20 years in the Bay Area, our head of engineering, etc., etc. So, I think there’s a dual element. We need to be quicker to think about the opportunities, and not immediately jump to all the risks and reasons why we can’t do it, or wait till someone else hazes are first. And then secondly, say, “No, we have some of the best capabilities here.” I mean, I’d love to see an “Own the Podium.” I love that—my parents live in Vancouver; I went to the Olympics in 2010—and like, why don’t we get that kind of, you know, maybe a little bit of strut, healthy arrogance, whatever you want to call it, that we are the best. Just like winning 14 golds in 2010.
Jordan Jacobs
So, okay, there’s a lot there. First of all…
Martin Kon
I’m not opinionated at all.
Jordan Jacobs
We have these conversations on text five, six times a day. Martin has no opinions. So, first of all, all of you have either come back to Canada after time away, or come here in the first place as immigrants, right? That’s across the board?
Tony Gaffney
M’hm (affirmative).
Chris Walker
Transplant. Yes.
Jordan Jacobs
Transplant, transplant, return…
Mara Lederman
After school.
Jordan Jacobs
…return. Okay. So, it is one of the big advantages of the country, is we have this amazing immigrant population, and also the proximity of the US, where people can go and work, and come back. We could pull people in who’ve experienced things in the US that can’t here. That doesn’t seem to have [indiscernible], though, the buyers, the customers for these companies. So, Mara, take us through your experience in selling into hospitals. Is the demand, is it harder to do that here? Do people understand what the opportunity is here?
Mara Lederman
So, let me make a couple general comments, picking up on what Martin said, and then we can talk about hospitals. The big question to ask is, like, if everybody thinks AI is so valuable—and I think we all do—why are companies not investing in a technology that is so valuable, and that should, in theory, kind of drive productivity and profitability? And the easy answer is, “Oh, Canadian companies are too risk-averse, or they’re not innovative.” But kind of, that’s like the one-line answer. And I think we need to unpack that question a little bit more. I think some of it is, they do think everything south of the border is shiny, and they’d rather do that. I think some of it is they’re risk-averse, and so, they go with, you know, kind of the big names. But I think there’s a couple of other things that are worth thinking about. We, I’m going to, I’m an economist, and I used to be an antitrust economist, so I can say this: we don’t have healthy competition in a whole lot of our industries. And I don’t think people understand the link between innovation and competition. What’s going to get a large organization to adopt a new technology? Competition. Shrinking margins, right? When you have competition, you know, what does new technology do? It makes your product better, so you can sell more and charge more, or it lowers your costs, right? If you don’t face competition, you don’t really need to do that, because you just keep passing on higher costs and higher prices. And so, people don’t often make that link. I think the link is very real.
We also don’t have that many large Canadian companies with global ambitions. So, what’s another reason to adopt technology? Because I don’t want to just be leading in Canada; I want to lead in the world. We don’t have that many. Our big, you know, we have big banks. Our banks, you know, from regulatory reasons, you know, some of them are, you know, in the US. But we don’t have a lot of huge kind of companies with global ambition. So, I think from a kind of ambitions and incentive reason, sort of structurally, it makes sense. It’s unfortunate, but it makes sense, that even in the private sector, we don’t adopt these things.
I think a second thing to unpack is kind of the attitudes around risk, right? If you adopt new technologies and you innovate, some of those projects will fail. And that was still the right decision, right? People need to get really—like, that’s the nature of experiment. That’s the nature, it’s what happens in VC, right? I mean, you do something truly on the frontier, and you do 10 of them, and some of them are going to fail, but it was still the right decision to do all 10 of them. And that, that’s the same reason I think we don’t have a lot of entrepreneurs in Canada. Like, that is a mindset that I think we’re not very comfortable with in Canada, and that we don’t give permission within companies to say, “Hey, we’re RBC, or whoever, and we worked with the startup, and a bunch of the things they did, we cancelled. That was still worth spending money on.” So, I think just like framing that, and figuring out—because I know we’re going to go to, like, what can government do? What do we need to do? We need to fix some of those things.
I think when it comes to healthcare, I think healthcare is, it is definitely easier to sell this and market it in the US. Not only the US is, of course, you know, a much bigger population, they spend 10 times per capita on health IT t than what we do. So, if their population is 10 times ours, their health spend is about a hundred times ours. Second thing that makes a difference is—you know, say what you will about their healthcare model versus ours, everybody likes to complain about the US, it’s private, 50 percent of healthcare is paid for by the government in the US, anyways—but it has very sharp incentives. And I will tell you, I have, every US hospital I’ve talked to cares more about quality than most of the Canadian hospitals I’ve talked to. Because if they don’t provide quality, it’s not just that they show up on the front page of the newspaper, right? Their contracts with their payers get cancelled. Massive, massive contracts, covering, you know, thousands, and hundreds of thousands of patients.
So, I think what we see in healthcare here mimics what we have in other industries. By design, we don’t have competition, right? We have very risk-averse attitudes, for good reason, because the area in which you’re putting a new technology is one where lives are on the line, as opposed to other things. And then, we have poor incentive models that come from the separation of the payer, in most cases, the province, and the adopter, which is typically the healthcare delivery organization—in our case, a hospital. So, one of the products we offer is a tool that predicts when someone’s ready for discharge. The idea is we can front-line, sort of, frontload some of that discharge planning, and reduce a couple, you know, some length of stay at the end of the stay—not while the patient is acutely sick, but at the end.
I’ve had hospital executives—so, when we say that, you know, “We can speed this up, and you can see more people, with the same number of beds and the same number of staff,” and they say, “Well, why would I want to do that? I don’t get paid for seeing more people,” and now you’re just taking a healthy person at the end of their stay and putting in a sicker person. Nobody wants that. I mean, the incentives are perverse. And the US, in many cases, have incentives that reward the benefits that innovation can drive directly.
Jordan Jacobs
Okay, so—and this is compounded by, historically, if you’re a second mover in technology, you let someone else go make the mistakes, spend the money, you know, kind of clear-cut the jungle, and you just follow on the nicely beaten path. AI is different though, right? AI is, if we think about AI as software, software traditionally is: you code it, you ship it, it doesn’t get any better until you ship the next version. So, all the users out there are using a static version of this thing that is not improving, whereas AI is learning software that is essentially getting better in real time, across the entire user base, and personalizing to the individual, in many cases. So, on two dimensions, the improvement is not just linear; it’s actually faster than that. So, the first mover is actually gaining an advantage every minute that this thing is deployed, and everyone else is falling further and further behind. Which is completely different from every other historical paradigm shift in technology, and adoption curve in technology. I don’t think that that is understood well enough. It is, certainly, by big organizations like Google and Microsoft, who have been at the forefront of spending enormous amounts of money and reaping enormous profits out of it—and Amazon, and Baidu, and a bunch of others. What can we do in Canada to help fix this?
So, I know one of the other things we need besides people is we need access to compute. Do you want to talk a little bit about that, Chris, and what you see—I mean, you can look at it from an international perspective, as well. It doesn’t have to be a Canadian-only perspective. But what is happening out there in the world, in terms of governments being willing to spend on compute?
Chris Walker
Yeah, I think one of—and just to pick up the thread from earlier—one of the things that’s different, I spent my career in semiconductors with Intel. A lot of time, you know, all my career has been Silicon Valley, Taiwan, Shenzhen. And what, the environment or what’s different about that isn’t just, you know, unbridled ambition and risk-taking—which, you know, we in the startup community, here, obviously, have in droves—but it’s a community. It is, to overuse the word, you know, ecosystem. They feed off each other, they support each other. And that’s a big differentiation about why they’re competitive at scale. This isn’t policy-based, it’s not, you know, just my backyard, it’s there’s a network effect to the technology, hardware and software, that makes, you know, the companies in Taiwan, the companies in Shenzhen better, because they do work together.
And I think, on the underlying compute, that’s absolutely the opportunity to do the model optimization with hardware together, in a kind of fellow traveller—because there needs to be differentiation to compete, right? To make your one solution faster, better based on that, you know, that exponential effect, because you’re learning. The more you’re co-optimizing, that just, you know, puts rocket fuel underneath it. And so, the underlying compute, the opportunity is to co-develop, co-design, and have, you know, a mutually supportive environment, whether it’s here in Canada, or Canada and other parts of the world. And that, I think, is a huge, huge lift to how Canadian companies will compete. The risk tolerance is there.
I’d say the other one—especially for our university friends in the back—it’s in your career, in how you approach your job landscape. Go in and out of, do a startup, do a big company, go back and forth. Because that cross-pollination is going to be what’s going to lift up risk tolerance, and things like that. In terms of government policy, you know, where it’s been most effective, what I’ve seen across the world, is not laissez-faire, but where it’s supportive, and lets people run. And in many cases, that’s, you know, creating the flywheel effect. You know, it’s the stuff that we see in terms of research, it’s actually being an open and welcoming climate for startups. The other aspect to it is, you know, create the pull. We talked about, you know, industrial segments, municipal segments, that can deploy AI in healthcare. Create a preference. Create a pull for solutions that have, you know—originated, designed by, coded by, in Canada. And that’s not handouts, that’s not, you know, big policy stuff. That can be done at a local level. But having that incentive base really helps pull through the underlying hardware and software technology, to allow us to compete at a global scale.
Jordan Jacobs
Tony—we’re gonna get to some questions from the audience in a second—but what specifically can the government do about compute, to make sure that Canadians have access? We’ve seen governments like the UK, France, Germany, Japan, Saudi Arabia, UAE literally buying up chips directly, as governments, which has never happened before. Explain just quickly why that is, and what Canada could be doing about that.
Tony Gaffney
Okay, sure, delighted to, Jordan. The, and the pace of what you just described can’t be overemphasized. The, you know, the UK moved in the past six months to stand up a whole hyperscale compute capability. So, in terms of what can the government do, I think the, I mentioned earlier, a national strategy. When it was released originally, infrastructure, compute specifically, wasn’t as key a component. With Gen AI, compared to 12 months ago, with more regular AI, the compute power that’s required is thousands of times more powerful—so, just giving you a sense of how quickly all of this is moving. Governments around the world are stepping in. Number one, to support setting up the compute infrastructure right away, in the context of having more of a national compute infrastructure to support AI over, you know, the near-to-medium long term. So, they’re stepping in to do that. They’re stepping in to work with research, startups, industry, to ensure that nationally, they’ve got the purchasing leverage to be at the table, and secure the supply now. They’re also looking at what it requires to stand that up. And it requires specific data centre capabilities, specific to the nature of the compute we’re talking about, and it needs to be supplied with green power, or we get all of the other complications that we’re talking about.
So, I would say that there is no need for the government in Canada to, you know, wonder what they should do. Best practice has emerged. Other countries are acting. For us, if you think about it, it’s a national supply chain risk. If we can’t get the compute we need—the first question students ask us or faculty ask us when they’re coming to Canada. It used to be, who am I going to work with? And we could share the story about Jeff, and who else, and other talent that were here, and they came. Now, the first two questions they ask, what compute can you provide me with, and what data is going to be available to me, to do my work? So, what’s at risk here is not just supply chain. What we’ve built, the research work that we continue to lead, the talent pool that we have here, it’s top of mind for them. And people either won’t come—and that’s really bad—people who are here will consider their options—that’s really bad—and the startup community that’s flourishing around them will also consider their options, as to where else they can go. And France has, definitely, an eye on attracting them.
So, I don’t have to continue the story. Investors have come here because of the talent pool. We just cannot let that happen. And first steps—and we are working closely with them—is to have the government step in, and follow emerging best practice, which is what I just described.
Jordan Jacobs
Okay. I’m just going to do a time check. How much—okay. So, we’re going to do a lightning round of audience questions.
Tony Gaffney
Okay.
QUESTION & ANSWER
Jordan Jacobs
There are a number of questions that kind of concern risks. One is, jobs being at risk, another is malicious activity. I think there’s worry about both existential risk, which has been talked about a lot in the press of [indiscernible], there’s also the risk of just bias in decision-making, bias in hiring as a result of AI being implemented, misinformation coming up in elections this year. How do we deal with this? Is, is—I’m just going to ask you each to kind of say 10 seconds worth: should we be regulating this, like, top-down? Should it be voluntary? Or should it be something else? Tony?
Tony Gaffney
I, I would say light regulation on top, to address issues like data, broadly, is important. But then, I think we need to focus on where AI is going to be applied, which is across all industry sectors. Pretty much every industry sector—whether it’s health, as well, and government have—have ways to deal with high-impact products and services, which AI is. And we should start there, and modify the, the regulation to ensure that it’s fit for AI, and take advantage of AI to achieve that, both from a regulatory and compliance perspective
Jordan Jacobs
Chris?
Chris Walker
I think it starts with transparency. So, have people aware and have the knowledge base to make choice, understand whether it might be bias or not, so they can understand, make their own decisions. I think it’s also important that we don’t broad stroke, you know, AI is, is everything. Because I think it’s important for the segments, the people who have the competency and knowledge by, you know, you don’t want to—you know, automotive should be, you know, people who understand transportation should do that, right? And so, not to slow down the positive effects, and the safety factors, and things like that that can be deployed today, by having big, unruly, boil-the-ocean policy, and regulation.
Mara Lederman
I agree with them, but I’ll say new things just to, to, to expand the conversation. When we think about risks, I think it’s helpful to think about those risks against the backdrop of what’s being done without the AI. So, I’ve heard someone say, “Well the algorithm’s only right 80 percent of the time. Should you report that to the patient, the 20 percent of the time the algorithm is wrong?” And I think, well, how many times is the doctor wrong? And how come we’ve never tried to be transparent about the number of times the alternative, the, the low tech solution has been, has been wrong? And so, we need to take these things seriously. But we can’t think that no other technology makes mistakes. You know, the COVID tests were wrong a lot of the time, too. We were okay with false positives and false negatives, there. And so, it’s not that different.
On jobs, we could have a whole conversation on jobs. But I’d love to see the conversation think about the industries where we have really pronounced labour shortages—healthcare being one, I think, good teachers being a second one. And let’s frame—I’m also worried about what ChatGPT is doing to how we teach—what if we could build a chatbot that was as good as the best grade five science teacher in the country, and gave every kid in the country access to that chatbot? That feels like that would be a great outcome, not one we should be scared of. So, let’s switch that conversation a bit.
Jordan Jacobs
Martin, you guys have been in the forefront of this with the UK government, EU, US, Canada. In a global competitive environment, where you’re selling around the world, is there a risk that, if Canada overregulates, it basically regulates out innovation, and pushes companies like yours away?
Martin Kon
A hundred percent. It’s a great place to be for many reasons, as we talked about. But if there are onerous regulations around copyright, or, or deployment, or risk, or whatever it is, we just move to the US, unfortunately. Now, the good news is, I think the administration here is very open-minded, and they also work closely with the US, the UK, and Minister Champagne, and Secretary Raimondo in the US, and Secretary Donnelly in the UK, are working together on this, and the G7, et cetera. So, we’re pretty, we’re pretty optimistic that things will move in the right direction. I think the, I think we’re all in agreement, a sectoral approach makes sense. We already have an FDA—or whatever the equivalent in Canada is, I’m sorry—we don’t need some, you know, Czar on top. You know, YouTube has dealt with misinformation for years. Now, it’s just a little bit more sophisticated. We’ll just have to get a bit better at doing it.
But one thing is clear, this is happening, period. And so, that people saying, “Oh, you know, should we allow it? Should we slow it?” It’s like, the horse is out of the barn; it’s, let’s figure out how to harness incredible potential. And it is extremely important to think about the risks, and think about things like bias and misinformation. I love what you said about the, you know, doctors being wrong sometimes. There’s been work done by Google, by others, that show that, actually, Google AI tools make better diagnoses than doctors do. There’s also been a lot of work done in financial services that these models—yes, there is some bias, but the bias about making loan decisions is substantially lower with AI than with humans. Humans are the most biased decision-makers ever. And again, no one really kind of calls them out on that.
Mara Lederman
Right. And you can audit it when it’s done by AI.
Martin Kon
So, I think we talk a lot about human-in-the-loop. And so, that’s, I think what regulators, policymakers should say, “You can’t prescribe drugs through a chatbot; make sure a pharmacist gets all the advantage of a thousand times more information and synthesis to make a really great decision,” and then they check it, and so on. Same with a loan. You can have a loan officer, you know, give her, you know, masses of information and support to make an incredibly well-founded loan decision, but make sure there’s a human in the loop, in that process. So, I think that it’s sort of a combination, in terms of what government can do, be a customer. I mean, you know, Canada keeps saying, you know, “Oh, we’re so great, Jeff Hinton, and so on.” And yes, we are. Yes, we are. The students at the back are a testament to that. But if you want to show that you’re innovative as a country, be a lead user. The G7 presidency was in Japan, and they did the Hiroshima thing, and they said a bunch of good stuff. It’s now in Italy, we’re working with them, they’re going to say some more stuff. Next year, it’s in Canada. So, we were in Ottawa last week, we said “We have 11 months, let’s build some stuff,” so that when the presidency is here, we’re not talking about it; we’re saying, “Oh, look at the way that we allow people to get information on renewing their OHIP card after 23 years, so they don’t have to wait five hours,” like I did. “How do we make it easier to get information about immigrating, tax policy?” Just easy stuff. And then citizens, trust it, say “Wow, this is amazing, this made my life better.” You know, doctors spend three-and-a-half hours a day typing up notes and discharge patients. Larry Ellison at Oracle used our technology, did a demo, he announced us at Oracle Earnings. He believes he can get that to 20 minutes. That’s three hours more time they can be operating, diagnosing, sleeping, training themselves, seeing patients, whatever. That’s not making jobs go away; that’s just meaning that humans focus on higher and higher order…
Mara Lederman
It’s productivity.
Martin Kon
…productivity, and…
Mara Lederman
It’s exactly productivity.
Martin Kon
…get rid of the low-value add stuff, and focus on the high-value add stuff. But the government can start by adopting this themselves, potentially with Canadian companies at the core, and showing that they, that they, that we are an AI-first nation, because we use this stuff.
Jordan Jacobs
Okay, I think with that, we’re out of time. Thank you, everyone, for joining us on the podium, and the audience.
Sal Rabbani
Thank you very much, Jordan, for facilitating today’s conversation, and to the greater panel. I’d now like to take this opportunity to welcome Nicole Foster, Director of AWS, to the podium for the appreciation remarks.
Note of Appreciation by Nicole Foster, Director, AWS
Longest job title ever! My name is Nicole Foster. I have a dual role with Amazon Web Services Public Policy, so I both oversee Canada, but I also have a global role in our AI policy strategy globally. I’m so delighted to be here today, and want to thank the panel so much for being here. And Angus, I’m delighted again to highlight the great research that you’ve done, that we were able to support. And I think it’s so important for us to kind of understand where we are, so that we know where we need to go.
AWS has a really strong presence in and commitment to Canada. We have two infrastructure regions here, both in Montréal and Calgary, to support our Canadian customers. And we have plans to invest up to 25 billion dollars in this digital infrastructure. Globally, over more than 100,000 organizations of all sizes are using AWS AI and machine learning services, and this includes Canadian startups, national newspapers, professional sports organizations, federally-regulated financial institutions, retailers, public institutions, and more.
Specifically, AWS offers a set of capabilities across three layers of the technology stack. So, we play a little bit of a role in all of the things that you heard a bit about today. At the bottom layer of the stack is our infrastructure, and we offer our own high-performance custom chips, as well as other computing options. In the middle layer, we provide the broadest selection of foundation models, including our Canadian champion, Cohere. So, this includes models such as Cohere, but also other providers such as Anthropic, AI 21 [indiscernible], and Stability AI, as well as Amazon’s own models. And then at the top layer of the stack, we offer generative AI applications and services.
We continually invest in the responsible development and deployment of AI, and we dedicate effort to help our customers innovate and implement necessary safeguards. Our efforts towards safe, secure, and responsible AI, are grounded in deep collaboration with the global community, including work to establish international technical standards.
We’re super excited about how AI will continue to grow and transform how we live and work. And I just wanted to highlight a couple of great takeaways from the conversation today. It’s such an important discussion, and I hope the Empire Club continues to host these discussions. But I think the comments about the need for sector-specific regulation are so important. We know that mitigating risk in AI is very use-case specific. And so, understanding how to regulate that is also going to be very sector and use-case specific. The importance of access to compute, it is definitely a constraint that we see around the world. We know that we’ve seen other countries also have adoption strategies that are targeting sectors where they know that they have labour shortages, so it’s a hugely important strategy for policymakers to think about. And of course, government adoption of AI, I think, will be to all of our benefit.
I just want to thank the panel so much for your great discussion, and we’re very pleased to support it, and thank you all for joining today. Thank you.
Concluding Remarks by Sal Rabbani
Thank you very much, Nicole, and thanks again to all our sponsors for their support, and everyone joining us in person or online. And I’d also like to thank The Logic, our media partner, who reports extensively on AI in Canada. This week, their newsroom published an ambitious and timely series entitled “Super Intelligence: Is Canada Ready for AI?” You can visit them at the back of the room for more information. As a club of record, all Empire Club of Canada events are available to watch and listen to on demand on our website. The recording of this event will be available shortly, and everyone registered will receive an email with the link.
Join us on February 28th for the Empire Club of Canada’s “Celebration of Black Excellence and History.” The panel of inspirational Black women discussing their careers and paths to success. Join us on Tuesday, March 5th, for an “Evening at the TIFF Bell Lightbox.” We’ll be hosting Sarada Peri, an expert communication strategist, and former speechwriter to President Barack Obama. She’ll help us understand the implications for AI, and for political campaigns, and democracy.
Thank you for your participation today and support. This meeting is now adjourned.
