Power Struggle
Improving the energy dialogue in Canada (and beyond) through honest, non-partisan, and fact- based conversations.
The energy conversation is personal: it’s in our homes, in our hands, and now, it’s in our ears. Power Struggle invites you to listen in on honest, non-partisan, and fact-based conversations between host Stewart Muir and the leaders and thinkers designing modern energy.
Watch videos at https://www.youtube.com/@PowerStrugglePod
Power Struggle
Can Canada lead the global AI revolution — or are we about to miss our moment?
Forget model tweaks—the real AI race is about power. We sit down with Kris Krug to map how GPUs translate into gigawatts, why data centers are the new factories, and what it would take for British Columbia and Canada to turn clean energy into sovereign intelligence instead of exporting value and importing electrons. From Vancouver’s buzzing AI meetups to Meta’s multi‑gigawatt builds, we connect the dots between policy, infrastructure, and the businesses already adopting AI far beyond chatbots.
Kris walks us through his path from creative tech to generative AI, the moment Midjourney reignited his practice, and the community that formed around practical workflows. We dig into the hard math of energy—firm power versus storage fantasies, LNG tradeoffs versus hydro flexibility—and the strategic fork in the road: build compute where clean energy is abundant or watch others do it faster with our resources. Along the way, we unpack how jobs are changing toward orchestration and token efficiency, why hallucinations fade with purpose‑built tools, and how young workers use AI as an operating layer, not a search engine.
We also spotlight a different kind of leadership: Indigenous‑led data centers in cooler, energy‑rich regions, sovereign data governance, and the Indigenomics Institute’s AI platform measuring economic value GDP misses. Alberta’s coordinated push (and Amii’s impact) shows what alignment can unlock; BC’s opportunity is to organize around clean, green, ethical AI that keeps models, insights, and benefits closer to home. If energy is the new cornerstone of intelligence, then policy is product design—and the decisions we make now will decide who owns the tokens of tomorrow.
If this conversation got you thinking, follow and share the show, leave a review with your biggest takeaway, and tell us: should clean power make AI a Canadian strategic priority?
The energy conversation is polarizing. But the reality is multidimensional. Get the full story with host Stewart Muir.
Reach out to us with thoughts, questions, or ideas at info@powerstruggle.ca
Linkedin
Instagram
Facebook
Twitter
🎧 For audio versions of our podcast visit powerstruggle.ca and listen on the go in your favourite podcast app!
Video available on Power Struggle’s YouTube! https://www.youtube.com/@PowerStrugglePod
In the last six months, every conversation about AI has gone from algorithms to energy usage. Um, that is the name of the game right now.
SPEAKER_01:Hi, I'm Stuart Muir. I'm the host of Power Struggle. Thanks for coming back. It's our new season, new studio. Great team here. I'm so happy to be back at it. Today my guest is Chris Krug. He's right here with me, and we're gonna have a fascinating discussion about AI. That's artificial intelligence and what it means in energy, but lots of other things too. Chris Krug, thanks for coming in.
SPEAKER_00:Hey, Stuart, thanks for having me in here. This studio is looking great. Tell me about the BC AI scene. Yeah, well, there is a lot of activity taking place in British Columbia and the act the artificial intelligence world right now. Uh we have a member-based nonprofit that people have been uh joining, and we've been getting together monthly uh at the Space Center. Have about 250 people who come out every month, and they represent a real broad cross-section of kind of the AI research labs, the AI entrepreneurs, and the kind of investors that support that market, as well as like enthusiasts and creative professionals and all sorts of technical innovation types, you know, and stuff. So it's it's a really awesome event and a booming scene. Are there any examples of companies, startups that are getting noticed beyond Vancouver that came out of here? Yeah. There's a company I'm tracking right now called Gumloop. They received$3.7 million last year, U.S. and ended up um moving down to San Francisco to kind of keep growing and stuff. Um I got my eye on them. Um Databricks is a huge San Francisco data company that's just uh announced last week the opening of an office here in Vancouver, a headquarters here in Vancouver. Um but yeah, there's a lot of interesting things going on in the in the scene, a lot of companies um getting funding, uh growing and partnering with other people and stuff. So yeah, we're we're seeing a lot of uh activity for sure.
SPEAKER_01:Aaron Ross Powell, you've been on the startup scene here for a long time in tech. What got you into what's happening now?
SPEAKER_00:Well, I mean, I've been a web guy since the late 90s, and I've been building websites and web apps all along the way. But when we sold the company Bright in 2007, I became the president of the acquiring company, which was Rain City Studios. And you know, that opened up uh an interesting decade of me kind of traveling around the world at tech events like South by Southwest. I was on the advisory board for almost 20 years and um talking about how artists and creative professionals can empower themselves through technology, essentially like disintermediate from old models with galleries and agents and other people representing their work and essentially connect directly with their audiences, their fans, build uh, you know, followings and movement, find their voice, find revenue streams that were new and stuff. And so I got to travel around the world, you know, doing that kind of stuff for almost a decade, um, which led me right up to the doorstep of the pandemic. And um I have a place on Galliano Island, and so I was uh doing some pandemic farming over there, and I had a visitor come by and he showed me Midjourney, which is an AI app for generated images, one of the kind of first cool AI apps for creative work, creative professionals. And he came and showed it to me in mid-2023, and it just kind of blew my mind. I I um had kind of made peace with my photography career as it were, kind of being the past pre-pandemic, you know, because you know it was a real lightning strike for me in terms of I got to photograph Richard Branson, uh, Barack Obama, Bill Clinton, like work for National Geographic and Rolling Stone and the TED Talk. So I had like the best clients in the world. It was a perpetual motion machine, but it came to a crashing halt during the pandemic, and I kind of just made peace that that had been the past, you know. But this guy came out and showed me this image generation thing, uh, Midjourney, and I was just like, I could see how it was gonna revolutionize the professional creative industries, design, photography, filmmaking, stuff like that. And so I just I went deep right away. I I got lucky. Uh I built a Discord server on the very first day I learned about Midjourney so that I could install the Midjourney bot on my Discord server and then just invite a couple friends to it, and we could kind of explore in private. Because the main Midjourney server was crazy. It was you know tens of thousands of people, and it it didn't feel like that safe of a space. There was hackers floating about and you know, crypto scams going on and stuff. And so I was like, let me just start my own Discord server, put Midjourney on there, invite a few of my old tech friends there, right? And uh, well, it was the moment really. Like things were right at the beginning of the upswing. And so a lot of these people either hadn't learned about generative AI yet or hadn't had a place with like collaborators where they could experiment, learn things, and like learn from each other, report back and stuff. And so my server just went astronomical right away. I had like a thousand people on there within three months, and they represented uh a 20 years of my colleagues and network and stuff. And so that was really kind of my entree into the whole AI world was mid-journey, into Discord, into connecting with all my old tech collaborators. And essentially, I started booking keynotes almost like right away. I started blogging about my learnings and my perspectives about AI. And um, like the Canadian Association of Architects and Engineers booked me to do an annual general meeting keynote talking about how uh creative professionals like architects were adopting AI inside their studios. And um really it just kind of went rocket ship at that point. You know, I just it I I started getting into high-level conversations about not just like what tools to use and stuff, but like what this actually means. That's that was what I was interested in. Like from the very beginning, it was like, oh, this is gonna change the professional creative industries.
SPEAKER_01:Right. Now you've been deep in it for a lot longer than most people. And one of the reasons it's so fascinating to me for power struggle is just the simple fact, and we'll come back to this, that AI requires an enormous amount of energy to run. But before we get into that, because I think it's the heart of a lot of the coming issues that we're gonna have to deal with, I'd I'd like to get a better understanding of where we are in the hype cycle versus the the long-term use case for AI. Because it seems like if you just judge by the headlines uh that that you see every day, you've got uh a bit of a sense that this isn't for real. But you've also got clear, compelling evidence that as we speak, there is a massive change going on in how businesses are adopting technologies, the number of people they're hiring, the kind of people they're hiring, the things they're doing. It's obviously real, but at the same time, it feels unreal. Do you know what I mean?
SPEAKER_00:Yeah. Um I mean you're right to to start talking about its power right away. Um in the last six months, every conversation about AI has gone from algorithms to energy usage. Um that is the name of the game right now. And uh I mean, if you look at just the uh orders that NVIDIA has for their H100s and H20s, their high-end GPUs for data centers and stuff like that. If they built and shipped all those, we'd need like 92 gigawatts of new power just to power them. Where's that gonna come from? It's not uh in the roadmap, you know? And so that's where you see people like Meta building this Hyperion five gigawatt data center and people firing up, you know, Three Mile Island and Microsoft making contracts for nuclear reactors and stuff, and um, everyone's talking about it right now. And I mean that's what makes this a really unique time for Canada in some ways, too, because you know, of our resources, our clean energy, you know, we um we have what the world needs in order to take us where they want to go in terms of a lot of this AI data center build-out and stuff like that.
SPEAKER_01:Aaron Powell So that's real. Now, I've been reading The Economist last week they had an article saying is the world's first unicorn by one person, a solopreneur, I think is the buzzword, around the corner. What do you think?
SPEAKER_00:Well, it seems inevitable because people are holding this up as such a holy grail and a target, you know? The idea that you could get to a billion-dollar company with no employees. Um I don't know how that really scales for society necessarily. Like if that's a really great idea for society, but I know that that is what people are endeavoring to do through. And I mean, we're seeing things close to it, like Cursor, uh AI agentic coding company sold recently, and I think they had like 30 employees when they sold for multiple billions, maybe$9 billion, something like that. I forget the exact number, but I mean, we're seeing things leading up to that. Um, I don't know, the hype thing is is pretty interesting. I think that um a lot of people have a lot of fears and concerns about the ways this can be rolled out in society, about who gets to roll it out. Like, do we trust the tech giants and governments to be the person that tells us what we need as it relates to AI? Are they truly looking after all of our interests or just those of the tech giant hyperscalers and stuff like that? So um, you know, there's a lot of, you know, I I just did some research through Angus Reed asking British Columbians their attitudes about AI, and 72% of people are more concerned than excited when it comes to British Columbians about AI. Um I want to get into your question, though, a little bit around like hype or whatever. Um, you know, AI is a lot more than just Chat GPT. And so, you know, with all most people's exposure to AI being primarily that, it's actually hard to understand a lot of the types of things that we're talking about here, you know, around like reinforcement learning or machine vision or some of the other more robust technical applications of AI. And those things are very real. Those things are very real. I mean, there's a lot of people generating immense data sets that are of very high value that the AI can be applied to to determine patterns, to make analysis of, strategic decisions and directions. So I think, you know, once you scratch beyond the surface of the kind of web browser consumer products that we're seeing from OpenAI and stuff like that, and starting into a lot of the other under-the-um hood infrastructure that really powers this revolution, that's where you see that we're, you know, AI is moving from um software development to essentially big industry. And happening rapidly. Um as rapidly as they can deploy it, actually. You know, I mean you see the project Stargate that Trump kicked off in America down in Texas, a$500 million build-out of data centers. Um there's like 6,000 people working 24-hour shifts on site there right now. You know, they're they're they're going as fast as they can.
SPEAKER_01:They are. And I'm kind of concerned that Canada is leaving itself out of the race. There is the scene you've told us about. It's not just Vancouver, it's Toronto, Montreal. There's some pretty cool companies happening. But at the same time, that massive investment of physical things to support AI isn't happening here. And it could. It could. And it's all driven by energy. And this really concerns me. I mean, you you see um this incredible phenomenon of say when crypto was was coming up, uh British Columbia uh uh had the opportunity, it had all this hydropower to be a place where crypto mining could take place at at scale. You've got a cooler climate, especially cool in some places. And uh uh being able to cool your data center is obviously one of the operational things if you can reduce the cost. Like it it made sense.
SPEAKER_00:Yeah, we got stable geology, you don't have to worry about an earthquake hitting your data center. That's right. You got always on power through hydro. Trevor Burrus, Jr.
SPEAKER_01:And not just hydro, but abundant natural gas here in BC, produced here, shipped through here. You could easily tap that. But it seems like there's been some legislative moves that have removed that opportunity. Uh regulations were changed to disallow crypto as something that you could use uh electricity for, and that still exists. So other data centers are coming up against this problem that they don't want you because you could be used for crypto.
SPEAKER_00:And then the other thing, which is that I mean that's a very interesting um example. I I view it as kind of positive in some ways because it shows that government policy can be used to set strategic direction when it comes to technology. And British Columbia, as I understand it, was trying to make sure there was enough uh electricity for EVs and for their heat pump initiatives. And so this is a uh a great example of government directing its power resources towards its strategic priorities. And so I think that we should make AI one of these strategic priorities for a bunch of good reasons, right? And we and we could uh set policy that would allow us to like thrive in those areas, especially around the building of out of this infrastructure. One of the reasons I kind of support the crypto decision was because it was um it was the old model. It was extraction. It was just taking our power and sending it off as you know, compute resources for crypto folks, you know, miners on the internet or whatever, um, without really returning too much of the value here. It's like sending out the logs or the raw bitumen from the olden days, and clearly that's not what Canada wants. We want to keep our stuff here and build true value and innovation in this economy. So, you know, like if we do decide to do some hydro-powered data centers, you know, for Canada's sovereign compute, keeping that here and actually using that intelligence to build things of value, to make medical breakthroughs on Canadian health data, or to, you know, really empower Indigenous nations to have a seat at the table when it comes to economic reconciliation. That can happen, I think, if we strategically build out this infrastructure in a way that aligns with British Columbians want here.
SPEAKER_01:But there's no way that will happen because we don't have the hydropower. We're importing power from the United States to keep the lights on in BC now. You know, you look at the Hyperion project in Louisiana for Meta, that's going to take about the same amount of power as you get from the new Site C dam in BC. The thing is that we have prohibited uh entrepreneurs in British Columbia from using the abundant natural gas we've got. Meanwhile, we are shipping our natural gas to the United States for a negative. It's actually not on some days, sometimes there's no revenue at all. And there's various reasons for that. And uh some of that is getting into the system so that Hyperion in Louisiana can produce this uh superpower level of compute for the United States, not for Canada. And probably if you're scrolling TikTok and using your your AI, you you might be getting some benefit from what's happening there with our gas that we've given away at a losing price. Um it's kind of staggering when you think about it. And it's all done out of uh sort of uh climate altruism, the belief that if we don't do it in BC, it won't get done anywhere, and that's climate leadership, but that's not the way the world is unfolding. It seems like a lost opportunity for BC. But in Alberta, they don't view it that way. They're actually building AI enterprise uh with their natural gas. So why is it that these two provinces are going in such radical different directions?
SPEAKER_00:Well, I think the jury's still out as it comes to LNG powering these data centers. Um I've seen some numbers that say looked at over 20 years, it's actually not any cleaner than coal. The methane in the life cycle of it isn't always accounted for. And we're seeing like these satellite images now of huge methane plumes that are coming from LNG processing, whether it's the liquefication or the transmission of it, that aren't accounted for in any of the numbers. And so I think that British Columbia's attitudes about whether we want to power the intelligence, innovation, economy on LNG, it seems to be like Jury might be out on that one still.
unknown:Trevor Burrus, Jr.
SPEAKER_01:Yeah. Well, I think in the Canadian uh case, our methane regulations have been exceptionally helpful in reducing and so much so that there's really no one producing natural gas more effectively than Canada is. And it's all in Western Canada. Um there are definitely places that provide the example you're referring to, but Canada is not one of them. Um think that uh because we've invested so much and so well in ensuring the the potential negatives of natural gas are managed, that we would say, now let's use it for something that we'd be good at. And instead, what we're seeing is uh the the U.S. is buying up uh pretty much all the gas turbines that you could use to run those centers as fast as they can be made. So it's even hard to get one of those if you wanted to.
SPEAKER_00:That's right. I watched a podcast last week from a guy who was talking about the failing grid infrastructure in the U.S. and in parts of Canada, and he was talking about how both like LNG and even some of the traditional other ways that they're powering these things is really not gonna cut it after about 10 more years, you know. And he's like, we're gonna have to, he was referring to like sci-fi. You see all these sci-fi movies, and like the planet is covered in solar panels, you know, and there's all sorts of electrification on the surface, and it looks like a mechanical planet. And he's like, Well, the reason it's like that in sci-fi is because that's what we're gonna need to do to get us where we want to go, essentially, or whatever. And so this guy's like, well, even the traditional sources aren't gonna be able to keep up. We need to do solar in mass scale on site and like have the data centers right there where the power is generated, and then ship the tokens out everywhere else.
SPEAKER_01:Yeah, and they have to find a way to get that solar 24 hours a day because they don't shut down at night. Yeah.
SPEAKER_00:And uh the guys are rolling out Yeah, they're rolling out huge battery trucks, man. In front of every meta data center, there is like a parking lot full of battery trucks. And those battery trucks altogether would give give you like 10 seconds of backup.
SPEAKER_01:So um to run things for eight hours or 12 hours of darkness is we're so far from that. And the amount of raw material.
SPEAKER_00:Well, and that's why BC has a strong advantage right now when it comes to firm power. And I would look at things like the hydro stuff. I think there's a lot of opportunities there to really like I think it that people want to be a part of the AI revolution and that their concerns will uh many of them would be mitigated by, you know, things like promising the best, you know, clean data token production out of our data centers in the whole wide world. I think people would come and um pay a premium for that, probably.
SPEAKER_01:Well, for that vision to happen, I think we it's a good one. We'll need to build some more big hydro because we're pretty much tapped out. Yeah. The electricity we're we're importing from the United States every day. Some of it's from solar in California, but a lot of it's from natural gas. Again, we've we've sent our natural gas to Washington State. They turn it into electrons. We import those back up here and call ourselves the climate heroes, which is the funny irony of the whole thing.
SPEAKER_00:And um, well, I think that's what I respect the most about your show, really, is like um centering power at the heart of this economic development and the future that we want to have is essential right now. It's like every highest level conversation in the whole wide world is how is about energy production right now. And we've got, you know, hydro here, we've got some pretty awesome geothermal type stuff that can be developed over time. There's some nuclear research going on, you know, there's uh Pacifican and other federal agencies have invested quite a bit in some of the um fusion operations here in in British Columbia and stuff. And so um developing a strong policy around energy and a plan and starting to get to it.
SPEAKER_01:Well, one of the things, I mean, it's almost like we're we're hitting the wall. You know, what do we do with this? Because it's clearly going to need more energy. The the projection when you uh look how much the U.S. is building, that'll be as as much as India uses as a country in you know five years' time. It's it's just a staggering amount. And there there is no ceiling on it. So where is it gonna come from? But it's not just that question. So uh how are we solving uh the other issues, the environmental issues? And it seems to me that AI ought to be one of the primary ways in which we can advance, accelerate human intelligence to be able to solve problems just like uh genomic sequencing has allowed these incredible medical pharmaceutical breakthroughs to occur. Do you think that's gonna happen and who's working on that?
SPEAKER_00:Well, it seems like that is um likely, but how well the benefits of that are distributed is the part that I have some you know questions about and stuff. So it's like, how do we ensure that if we're generating a bunch of energy up here to power data centers, that the tokens that come out of that that are used to fuel innovation and breakthroughs and stuff, how do we make sure that that kind of gets back into the hands of Canadians as well? And um, you know, there's some pretty far out ideas when it comes to that. Sam Aldwin was talking a couple weeks ago about um he was taking some swings at universal basic income and even like the idea that like maybe we're all gonna be out of work and lacking purpose someday in the future. And he was like, he's like, let's not talk about UBI, let's talk about maximum global wealth. And he's like, it sounds preposterous, but here's what I mean. And he was talking about how in the future he sees we're gonna be measuring countries' output, not by GDP, but by tokens. And he's like, and he he right recognizes that people like himself, who own all the infrastructure, the data centers and stuff, stand to benefit the most from this also. So it's like, how do you align the needs of society and individuals who don't have a skin in that game with the needs of the people that uh own and regulate and finance all that kind of stuff? And so his his thought experiment was like, okay, let's say uh US and Canada produce 20 trillion tokens a year. He's like, well, maybe we go with like a 60-40 split on that, where the people who produce them, the data centers and whatever, get 60% of those tokens to use for whatever they want. But then the other 40% are allocated to society on an individual basis. So like you and I would each get a million tokens a day or something like that. And he's like, for people that need UBI, they can sell their tokens to the next guy and take the cash. But for those who are looking for access to innovation infrastructure to develop their own ideas to make startups that compete with these big incumbents, maybe they go around pulling uh other people's tokens that they can, you know, get access to or whatever. And it's like, in this way, he feels like that the you know the growth of uh the hyperscalers and the powers that be that own the infrastructure and the um inclusion of the rest of the sci society could go hand in hand.
SPEAKER_01:Trevor Burrus, Jr. Yeah, sounds a bit like a sci-fi novel. I'm not quite ready for that concept, but someone's out there talking about it. It's a new idea, interesting.
SPEAKER_00:Um Yeah, and that's fair, man. That you you need to be focused on short-term policy, politics, industry type stuff here. And then we got lots of other folks, you know, looking farther down the road.
SPEAKER_01:And I think that's that's well are those ones who will just have all the money by then anyways, and they'll just do what they want. Um is that is that the way we're headed?
SPEAKER_00:Aaron Powell It seems like um unless we put some stop gaps or some guardrails, people like to say, in place, it does seem like that's the future I see. I mean, it's it it feels like there more and more it's gonna go in the hands of a smaller and smaller uh amount of people.
SPEAKER_01:Aaron Powell And it's already pretty concentrated wealth. Yeah. That that sounds like we could be getting into a dystopian potential future of this.
SPEAKER_00:But uh I think that ultimately the prospects are pretty positive. But I think for society on a whole, in terms of progress around, you know, health, medicine, education, I think that the uh the future is very bright for these things. However, we're gonna be in for a bit of a ride. I think we got probably 20 or 30 years where everything changes from the ground up, where all systems are in flux, you know, all organizational structures, whether it be government or voting or education or taxation, all these things are gonna be rethought again in the next short period of time.
SPEAKER_01:Well, I've been part of a lot of technology transitions just in my own career and in my life as a journalist, you know. The first typewriter I ever used was manual. I don't know if everyone even knows what that what that means. Manual typewriter, what's he talking about? I did take um typing in high school, not keyboarding. Same here. And then I got an electric typewriter. Yeah. And then after that, I got the first uh pretty crude word processor. It had one line on the screen, a little LED screen. You're an early adopter. Then after that, you know, actually a proper screen, and then uh I do remember the day I I had worked at I was at a magazine in Hong Kong, and we used to communicate with our offices around the world using telex. So you have to learn telex ease. It's a very refined language, owing to the fact that it's about a dollar a character. So you gotta make your telex really short if you're gonna say anything, you know, in a few lines. And so that's uh that's a language that was evolved over the history of the telegraph. And um no sooner than I did I learned that than one day they rolled in this new thing. They called it a facsimile device. And what's that? Well, you can print out your telex and put it on the fax. And for the first few weeks, we were all using telexies to write out our message to Hamburg or uh LA. And then one day I think it dawned on us all at once, wait a minute, maybe we can just write this in real language and then print it out from the computer and and put it on the fax, and we don't have to use it that way.
SPEAKER_00:Anyways, that was that was so that was how you did text messaging. How did you do Instagram?
SPEAKER_01:How did we do Instagram? Yeah, uh exactly. We had codachrome, it went to the lab. But um, you know, those technology transitions were not that disruptive. There were ones in, say, the newspaper business where there were labor uh issues because uh those with certain skills were suddenly kicked to the curb. And obviously when that power shift happens, uh you're gonna have that disruption. But what what we're looking at now is to me both uh a disruption, but also this massive opportunity, and I'm seeing it personally in how I use AI as an opportunity to expand my repertoire to do things faster, to have not just a better tool, but uh an assistant or assistants that are letting me move faster. Do you think that this is the way most people are experiencing? Or you know, I get I get a sense even younger people are afraid of this tool that's gonna take away their tools on the job. Where do you think we are?
SPEAKER_00:Well, um with that Angus Reed research that I did um that we just got back in the last couple weeks, uh definitely the younger you are, the more optimistic you are about it. And the higher your adoption is. And it's it's very clear. It's like 70% in your 20s and going down to like 20% in your 80s in terms of adoption of AI and usage of it. So people younger folks are more, and younger folks are using it different ways than older folks. So um again, I I quote Sam Altman. He's like, I I like looking over people's shoulders to see how they're using AI, and you know, he's for the most part, he's like folks over 60 are using it as a Google replacement, a question and answer machine. Um people in their, you know, between 25 and 60, they're kind of using it for general life coach. And so uh, in fact, I saw data yesterday that um uh uh counseling and personal relationships were like the number one, it was like MIT studies saying the number one and number two usage of these things is like for counseling and and psychology type stuff. People are people are talking to it like it's their coach, you know, and stuff. Um so that's what you see in the middle group. And he was saying the bottom end that uh the youths are using it to tie together all other systems on the internet, like essentially an AI operating system that interacts between all their other things. So I find that I find that pretty fascinating. Like I think the ways that we use it um are still sort of emerging. That was one of the first things about ChatGPT that really like switched me on about it was I heard about it from like five different people before I used it, and all of them were non-technologists, and they were all using it for different things. Like one guy was a mechanic who was referencing auto parts and different auto uh fixing strategies. Another was a guy translating like an Italian esoteric alchemy book or something like that. Another one was like my uh the the girl who I was dating's father, who's a construction guy who was using it to write back and forth his letters to the um to the client each day and stuff. You know, he could just dictate messages into his phone and have it send reports and stuff. So it's fascinating. I don't know if we know even know how we're gonna be using it all the way to the Aaron.
SPEAKER_01:No, if if youth, as you say, are these enthusiastic early adopters, let's not forget it's also youth who are gonna be going into the job market or are entering it now, and they're finding a totally different landscape in some occupations more than others. What what what do you think that is gonna mean in terms of societal attitudes towards AI and the impact?
SPEAKER_00:Aaron Powell Well, I mean, I think it's pretty clear that a bunch of jobs uh that we're currently familiar with are gonna be replaced or changed. What's less clear is how many new jobs are gonna come out the front end, you know? So if I look around Vancouver, I look at somewhere like Olympic Village, everyone who lives in those towers is like a video game designer, a UI or UX engineer, got an AI company, they're somehow involved with this uh they have a job description that didn't even exist 30 years ago. When I graduated from college, none of the jobs that most of my friends have even existed. So it's conceivable that just beyond being a prompt engineer or some silly thing like that, that there's gonna be a bunch of new jobs in the future that uh rely on AI. And I think there will be a lot of jobs that change quite a bit. One of the things we're hearing a lot about is we're all gonna become orchestration engineers. And like, what does that mean? Well, uh in my company, I've got 10 people, I give all 10 of them a million tokens a day, and they're supposed to use those as efficiently as possible to grow value in the company. And then I maybe have some extra tokens, and I can uh align those tokens with the things that are gaining the most traction or something like that. And so it's almost like a resource video game. You know, these resource management video games, it's like how do you get strategies developed efficiently for low-cost tokens where you're using models that are just as good but not too good so that you can get the cost down. And um and so I I think a lot of jobs are gonna be like this. They're gonna essentially be like playing video games where you're you're you're like trying to see how much value you can create through the tokens that you're allocated in the time that you've been given.
SPEAKER_01:Interesting challenge. So that's not uh the same as getting a law degree, going to a firm, working really hard for um a decade or so, and then moving up, um, probably the That is a thing of the past.
SPEAKER_00:Yeah, I mean, having deep encyclopedic knowledge of tort law or some sort of contract negotiations does seem like it's going to be less of a useful specialization in the future. Again, yes, not the first time, I guess, in this conversation I've heard referred to sci-fi, but like I often think of uh and even just visualize like the matrix in my day-to-day life now, where at the beginning he's with Morpheus and he's like learning. He had just taken the pill or whatever, and he's trying to figure out life in the outside of his little uh pod, you know, and and he's able to download knowledge like right as he needs it. Right in the moment that he decides he needs something, he's able to learn something, you know, about what he needs. And and it this seems kind of like the way of the future in some ways. So it's like I have this show on CBC about AI, and I do a regular segment, but sometimes they call me out of the blue and are like, hey, can you come on in one hour and talk about the work of Jeffrey Hinton, Nobel Prize uh professor in AI, doing neural networks and stuff? And I I have a generalized knowledge of AI, I know a lot of different stuff, but I don't know about backpropagation and neural and some of the innovations that Jeffrey Hinton made that resulted in his Nobel Prize. Well, I did a couple searches, threw them into a notebook LM AI, had it replayed back to me on a podcast, hopped on the bus, and after two listens to this podcast that was made to brief me to be able to speak about this stuff, I'm able to speak, you know, could connect the new knowledge that I just gained with all the kind of generalized knowledge that I just had that put me in that seat to begin with. And it it feels like downloading just in time expertise when you need it.
SPEAKER_01:But what if there's one crucial fact in this that is a hallucination? How do you know it's not there?
SPEAKER_00:Yeah, well, you that's looking at it from that perspective that I was telling you about of the of the web-based chat bots, the chat GPT, the consumer products, or whatever. So, first of all, yeah, those AIs are prone to hallucinations. They try to synthesize information from diverse disciplines into new ways and stuff like that. And sometimes they're just making shit up, and that's true. And so they've, on that consumer level, built other products like Notebook LM. Notebook LM uses AI on only five or eight documents that you give it. It's not gonna be pulling in stuff from unknown sources. It cites uh in its outputs where in your documents specifically it got that kind of stuff from. And so uh a lot in all the deep research models that have been coming out, like in the last six months, these are the thinking models, the reasoning models, they are also built to eradicate hallucinations through citations and other things like that. So that's just all on like this consumer level. But then if you get like a purpose-built AI where you're training your own model on something and then you're you know doing your own prompting and kind of rag environment on the top, you can very easily build tools that don't hallucinate. The hallucination artifact is kind of something that we've become familiar with through these uh general purpose, web-based ChatGPT-like experiences.
SPEAKER_01:Aaron Powell So as the as the product is refined for different applications, it's gonna get better.
SPEAKER_00:And people will see that. That's right. And also like, I can ask uh AI a question that's very prone to receive hallucination responses. And if I would have just asked a different AI that same question or asked my question in a different way, it wouldn't have been maybe so inclined. So already I can like recognize I I do a lot of like training around AI and I work with a lot of people as we bring up their skills and stuff. And um, you know, just I've I've started to notice ways that we can talk to these things that uh kind of incite more probability of hallucination than than um than it needs to.
SPEAKER_01:And it seems like everyone is finding their way to mitigate what they believe are the the the issues they might run across. No matter who the user is, I find there's all kinds of workarounds it. People are just evolving as users, they're not experts, they're just trying to make the best of it, make sure they're not embarrassed professionally if they go out there on something and oh wait a minute. That that left me stratd because it was a hallucination. Yeah. And imagine if you're a lawyer or someone uh who's uh who's uh got human lives uh at at risk because of their work. Uh you don't want to make these mistakes. So I I would imagine there is probably still a degree of reluctance. But from what you're saying, the the refined professional tools are there and they will become uh more understood over time, no doubt, eh? Yeah. Yeah. Um Well you know, we're talking about BC. We're sitting here in Vancouver, Chris. Um I just want to look at the Canadian scene because I know there's a few other things happening. We have a federal government with an AI minister. Right. That's a first. Minister Evan Solomon. Yeah. So where do you see this happening? Does does Canada have a chance to be globally significant as an AI player, or are we just going to be kind of one of the pack?
SPEAKER_00:Well, I think that we have great research labs already. We've got, you know, Yoshia Bengio who holds down the Mila lab out there in Montreal. We've got Jeffrey Hinton in Toronto who just won the Nobel Prize, as I mentioned. But then here locally, too, you know, we've got Alan Mackworth out at UBC who runs CADA, which is the Canadian Artificial Intelligence Association, and they've got a really solid, you know, 30-year lab out there. We've got guys like Steve DePolo and Philippe Pasquay, who both run labs at um SFU's School of Interactive Arts and Technology. Um, again, these guys are uh world leaders in their AI research and fields. So as the conversation has shifted to infrastructure and power, I think that our best chance to take all Canadians along on that journey and really to have a distinct and powerful product to sell the world is aligning around power and AI in the future. So, how do we build the infrastructure, clean infrastructure, which is what I hear Canadians want to power these things, and then not just ship them off like logs or bitumen, but to actually use those tokens to make cool shit here in Canada.
SPEAKER_01:The potential for British Columbia for Canada to be involved in that, I think, is is clearly becoming evident to some. One example I think of, I heard about recently, the Profit River First Nation. It's a small First Nation in Northeast British Columbia. They're located where most of the natural gas is. And they've said, let's build a data center here that will serve AI and whatever other needs come along. Yeah. Why do it there? Well, you might do it there because cooler temperatures, so less expensive to cool it. It's the ambient air that keeps it cool. Also because low-cost energy, they have that abundant mix of energy inputs. Natural gas is the most obvious one because it is reliable. Whenever you want it on, it's on. And it it's abundant. It's the the First Nations finally flexing and realizing they have the rights to be economic players and finding these opportunities. It was quite intriguing that when that was announced.
SPEAKER_00:That's a great example. I mean, I think that that, you know, this represents a time for us to truly leave our kind of colonial past behind and to build a non-extractive industry in AI. And so, like an extra extractive industry would be just shipping off our power somewhere else to be used as tokens to make breakthroughs and stuff. But instead, we could uh uh keep that here and and truly kind of like walk the walk of reconciliation that we've been talking about. And I think that there's some, you know, some other kind of history that it'd be nice to shed along the way. It's like we've been great at speculatively finding things in the ground and then you know selling that hype through junior mining companies and the whole stock markets that sprung up here. And a lot of times it was like how to find stuff, extract it, and flip it. But instead, if we can kind of close that loop, build truly circular economy stuff here, we can probably move beyond the kind of more dark underbelly of the investment climate that's been in British Columbia for a while and move into something that is really looking to be uh a leader to keep that stuff here and to make sure that the the outcomes of it serve all Canadians.
SPEAKER_01:Yeah, I could kind of wonder, you know, the difference in character between Alberta and British Columbia. In Alberta, they're pursuing AI with real vigor. They've got a lot of companies in Calgary that are affiliated with oil and gas because they're trying to optimize the the the data they have. There's huge amounts of data just like in mining and other fields, and you've got all kinds of companies that are that are coming out. Kind of different corporate history. You're kind of putting your finger on something in Vancouver. It still does have that reputation that was earned over quite a few decades.
SPEAKER_00:I'd like to give them some kudos as well. Um they were able to align as an industry, an AI industry, and lobby the federal government, and they receive, you know,$400 million or whatever to build that AMI institute. So I I respect Alberta's ability to get together as an industry and lobby the federal government programs to receive some of that Justin Trudeau,$3 billion of AI funding, and they use that to build the AMI center there. I've heard that the economic impact of that has been about$2 billion. So, you know, the feds put in$400, and then clusters start forming, all of a sudden, uh oil and gas companies become data companies and software companies, and so that's what we're seeing in Alberta. And that's because they've been able to get together, form um, you know, one one voice, one choir, and then sing their song to the federal government. And then that's very much what I'm trying to do with the BC AI ecosystem. I want to make sure that when these other big swaths of federal dollars are handed out, that British Columbia isn't overlooked again, that we can join the ranks of Vector Institute, Mila, and Amy across Canada as having a center for excellence here in AI, and that clean power can really be a part of that story. Trevor Burrus, Jr.
SPEAKER_01:What would you like to happen? What decisions are needed to get there?
SPEAKER_00:I actually think the responsibility in many ways is on us in industry to come together, organize our power politically, figure out what we need and want and are asking for, and then present a de-risk cohesive story about what we are good at here and the impact that we can make. And that is, in many cases, AI for the creative industries, AI for applied industry, the types of things that you're interested in. And then, like, there's a lot of ethical AI research that's being done here as well through Wask Center and other places around. So we can really be a leader in clean, green, ethical AI across Canada and the world.
SPEAKER_01:Chris, you're involved in something indigenomics. Tell us about it.
SPEAKER_00:Yeah, I'm the CTO at the Indigenomics Institute. I work with Carol Ann Hilton, who's like one of the leading indigenous economists in Canada. And she's written a couple books about uh economic reconciliation in Canada and Indigenous folks taking their seat at the economic table. And um, yeah, I've been working with her to build a indigenous indigenomics AI platform, an AI platform that um analyzes economic data from indigenous economic development corporations, from public sources, from um through our own research, and then surfaces the um stories and finances around um economic reconciliation.
SPEAKER_01:That's really exciting. Can you give me an example of what that might look like?
SPEAKER_00:Caroline has been using this indigenomics AI platform that I provided her to look for indigenous economic activity that is not accounted for by traditional metrics like GDP, but that represents significant contribution to Canada's economy. And we have an event on Vay Street later this year in just a few months where we're gonna be rolling out some pretty shocking numbers as it relates to indigenous economic activity that is already taking place in Canada, that is not covered by GDP, that's been unearthed through this process of um the Indigenomics AI platform.
SPEAKER_01:Well, I'd love to know more. In fact, I'd love to get Carol Ann onto our stage at IPSS, the Indigenous Partnerships Success Showcase, because we found that getting the data to tell that story can be really tough. So if you're bringing it forward, I think it's about time.
SPEAKER_00:Indigenous Data Commons, which is an ability for economic development corporations and Indigenous nations to contribute their data into a um sovereign data governance type format where we can use the platform to analyze the data and then share the analysis back with folks who contributed to the commons. Cool.
SPEAKER_01:Well, look, Chris Krug, it's been great having you here at Power Struggle. I think there's some fascinating topics. We've begun to delve into them. There's a lot more we need to learn, especially on the future of energy and how it fits into the AI revolution.
SPEAKER_00:It feels like the most important thing facing us right now, and I really appreciate you bringing me here and hosting this dialogue and really kind of continuing it out there in the world because if we can unlock how we can provide this clean, abundant AI resources and infrastructure to the world, we'll really be doing a good favor for Canada.
SPEAKER_01:That's right. This has been Stuart Muir at Power Struggle with our guest, Chris Krug. Thanks for joining us. Thank you.
