S6 E6 - Kendra MacDonald – Navigating Innovation Across the Digital Ocean

I have a conversation with Kendra MacDonald in this episode. As CEO of Canada’s Ocean Supercluster, Kendra is helping reshape how industries, governments, researchers, and technologists collaborate around one of the planet’s most important and least understood systems: the ocean. We dive deep into how emerging technologies like AI, autonomous systems, and quantum computing are transforming the ocean economy, while also exploring the organizational and leadership challenges that come with innovation at scale.

Kendra explains how the Ocean Supercluster was created as a national effort to accelerate commercialization, strengthen collaboration, and modernize ocean industries through technology. The conversation expands beyond marine innovation into broader themes of systems thinking, digital transformation, and the importance of solving business problems before deploying technology. Drawing from her background at Deloitte and her work advising organizations on technology adoption, Kendra shares why so many AI initiatives struggle, not because of the tools themselves, but because organizations fail to align culture, governance, and operational goals.

We also explore the tension between innovation and governance, challenging the common belief that controls slow progress. Kendra argues that thoughtful governance, risk management, and cross-functional collaboration actually accelerate innovation by creating trust and reducing downstream failures. The discussion touches on AI risk in high-consequence environments, autonomous shipping, data trust, digital twins, predictive maintenance, and the growing role of AI in optimizing maritime operations.

One of the most compelling parts of the conversation centers on the future of quantum computing and its potential impact on ocean science, logistics, climate prediction, and biodiversity discovery. Kendra and John reflect on how industries can prepare now for technologies that may radically reshape optimization and decision-making in the coming decade.

The episode ultimately becomes a broader reflection on interconnected systems, oceans, organizations, technology ecosystems, and society itself. Kendra offers a powerful reminder that the ocean is not simply a coastal issue, but a global infrastructure system affecting supply chains, climate resilience, communications, energy, and food security. It’s a thoughtful discussion about innovation stewardship, systems thinking, and the responsibility of leaders to build technology strategies grounded in purpose and long-term impact.

Transcript

[00:00:00] 

John Willis: This is John Willis again. We got another podcast, and this should be really interesting. I

think you'll... I think you're gonna enjoy this. And why don't we just start off with Kendra, why don't you go ahead and introduce yourself? 

Kendra MacDonald: Yeah. So Kendra MacDonald. I'm the CEO of Canada's Ocean Supercluster.

The Ocean Supercluster is a program in Canada. It's part of the Global Innovation Cluster program, and it's really trying to accelerate industry investment in the commercialization of technology or in innovation. My background prior to that, I spent almost 25 years at Deloitte, 12 years as a partner.

I worked in a number of offices all around the world, so Ottawa, Montreal, but also Australia Hong Kong, and then I'm now based on the East Coast. I'm based in St. John's, Newfoundland, and really have really worked hard to help companies understand the importance of investing in technology and increasing productivity.

John Willis: Great. And you talked a little about the Ocean Supercluster I guess sometimes referred to OCS, and I'll [00:01:00] probably stick with that from here on in, make it simpler for me. But so what i- what is this? What is a sort of a broader, like

the kind of people listen to this show they're boundary spanners, and they like to learn from other people who have experience in things like how do you do organizational dynamics?

How do you deal with, feedback or complex systems or all those things? So ques- it's a twofold question. What is the Canada Ocean Supercluster in a little more detail? And then then we can dive into what, w- what are the ways that you're learning or cultivating this this supercluster that can be helpful for, us boundary spanners to understand things, you know- Learn from.

Does that make sense? 

Kendra MacDonald: Sure. So OSC, if so we're gonna go with the acronym, so OSC. And we actually set out to change the way ocean business is done. So this was actually a huge cultural experiment across the country. If you think about the ocean, 70% of the planet, [00:02:00] and as much as we talk about various oceans, they're all interconnected.

So at the end of the day, you have this huge global interconnected system, and so how we operate arou- across the country matters, and then how we operate and coordinate across the world matters. So it was this it was a significant project in terms of how we operate. So for Canada's Ocean Supercluster, we are a public-private partnership.

And so we help de-risk innovation funding and accelerate commercialization. But if you think about where ocean is heading or the ocean economy is heading, so at a high level why this matters is because the ocean economy is doubling in size by 2030, so it's a $3 trillion US economy by 2030, so it is very significant.

90%, 85, 90% of our goods, so our supply chain, the majority of our supply chain, massive interconnected system actually happens by ocean. The bulk of our internet traffic that goes [00:03:00] internationally is by sub-sea cables. And then if you look at food, a significant number of our countries and a good chunk of our population gets their food, and that is the fastest growing source of sustainable protein.

A-animal protein actually comes from the ocean as well. So you have these really big systems, I would say, and then you're seeing again on the energy side we have existing offshore energy, and you are looking at a significant increase in renewable energy. So these massive sectors that all touch on the ocean.

And so now what you are also seeing is you're seeing instrumentation of the ocean. And so whereas we couldn't always get everywhere, now we have the technology to be able to go everywhere and anywhere, deeper and farther than ever before. And so you've got this set of sensors, buoys, autonomous platforms.

You've got drones, you've got satellites. Now you've got low Earth orbit [00:04:00] satellites. And so this is creating this massive interconnected system and instrumentation of the ocean that is producing a bunch of data, how we share that data so that we can optimize our use of the ocean as an asset, but also protect the ocean because we're seeing extreme weather patterns, we're seeing sea level rise, all the things that come with a warming ocean and a significant change in the environment 

John Willis: Wow.

Kendra MacDonald: So hopefully that helps. I feel like that was a long answer, but hopefully that was helpful. 

John Willis: That was awesome, right? No, because it's been a who'd have thunk 70% of the earth would give us a wealth of knowledge about how we live. Yeah. That's pretty good. I think that one thing I would quickly ask, which is a little bit off the, the beaten path, but, and not, I don't wanna get political, but there's a lot of frustration for people like me, and probably a fair amount of people who listen, about innovation in the US right now.

And Canada's always been pretty, [00:05:00] one thing I've always admired about Canada, like the comedians, the innovation. And is there anything like this going on in the US, or is there a counterpart or are we stifled, or is there some things we, resources we could look at very similar projects going on in the US?

Kendra MacDonald: So there are actually, there are a number of clusters. So one of our roles right now is we co-chair what's called the Blue Tech Cluster Alliance. So that actually includes TMA Blue Tech, which is a cluster in California. 

John Willis: Okay. 

Kendra MacDonald: There is also... they're more state level. Okay. So Washington Blue is also a cluster that exists in the state of Washington.

You've also got Alta C, which is in the Northeast. You've got Rhode Island, which has a lot of activity. I can't remember what their cluster's called. I think Houston has one as well. 

John Willis: Okay. 

Kendra MacDonald: And so it really, there are a number of these blue clusters that are really trying to, again, we're at a national scale, but bringing together, academia, government large industry, small industry, and investors to be able to work together to accelerate the development of these solutions.

John Willis: Like I said, I always let [00:06:00] these sort of podcasts go. I have like my template of what it's gonna go by, but I guess the question now, how, what got what sort of drove you into this? So you, it sounds like you were doing institutional, Deloitte and Touche like stuff, and then all of a sudden this pivot to something.

It seems like there's a passion that was inflamed or cr- created a flame or something. What sort of drove you to this? 

Kendra MacDonald: Yeah. So I was lucky. So I'm an auditor and an accountant by background, but I was lucky from very early in my career, I started as a co-op student. I worked a lot with telecom companies, and I really liked tele- so I got to go see a satellite before it got launched and do a bunch of work on this, again, the system in terms of switch to bill, which at that time you paid by the minute.

So you had to understand the time of day, you had to understand where it originated, where it terminated, all the places that it transited, and then you had to try to share, revenue across all of that. And there was a lot of technology change that happened. The cable companies came into it. You had the voiceover IP, the internet came along without saying exactly when I started my [00:07:00] career.

And so what made that exciting was that I could understand the purpose of the technology. The purpose was to get, voice from point A to point B so that you could make a phone call, and everything in between mobile came in there as well, all the technologies in between.

Ultimately, that's what they were trying to do. And so I really enjoyed that part of the tech sector, and so I spent a number of years in there. And then about a decade or so ago Deloitte did actually a lot of work in Canada around productivity and the importance of investing in technology for companies, but also this whole set of emerging technologies.

So artificial intelligence, you had robotics, you had additive manufacturing or 3D printing. You had the gig economy or, the change in work on- workers on demand, which was coming, as well as cryptocurrency and blockchain. And yet there were so many companies that really weren't aware of the technologies.

It would be hard today to think that nobody's aware of AI, but a decade ago it was still, an [00:08:00] early stage technology. And I started working with companies on helping them understand digitalization, what that meant across a number of sectors from media to transportation, et cetera. And so when this role came along, it was an opportunity to bring these technologies into the ocean economy.

I think we tend to think of ocean, we think of marine biologists, we think of whales, we think of, dolphins, and all those things are incredibly important. But there is this huge As I talked about instrumentation, over the last five years we have gone from about 7% of the ocean floor in high-fidelity mapping to about 25%, but that means there's 75% to come.

So we're just in this really interesting moment as we see more and more technology coming into the ocean. 

John Willis: Brilliant. Yeah, so I wanna anchor that question. There's a question I wanna ask, but I don't want you to answer it right now, which is, and I won't forget 'cause I wrote it down, but just in case, which is like what do you see now versus then?

But I [00:09:00] think the thing I-- another area I thought that I was really interested in, and it's something I'm interested in my day job, which is you call it innovation steward- stewardship or innovation versus innovation theater or purpose-driven technology. And and we can dive...

I love a lot of my work over the years is trying to study historically things that have taught us organizational dynamics. I go back to people like Dr. Deming. I wrote a book about him. And then there's all these sort of loosely speaking management science, but like it's a lot more than that.

It's really organizational dynamics. So I'm incredibly interested in your thoughts about what as sort of innovation issues. So one last thing I'll say is right now I think and I wanna cover what you think about innovation in general, and then we can talk about where it might be applying to new technologies like AI.

But right now, I do think one of the inhibits to AI success in large organizations is this abil- not ability not to understand the gap between talent and [00:10:00] innovation, and really understanding what your organizational dynamics look like. So you think, tell the world we have to do AI, and you haven't really discovered what your talent is or how horizontal spread your talent is.

Anyway, so with all that, just your general thoughts about in, innovation stewardship versus theater versus, purpose-driven technology. It's a 

Kendra MacDonald: lot. Yeah. From, I, where that really came from is as I look at how The word innovation gets used, and it gets used for a lot of things, and it's not meant to be just technology innovation.

It's really, if you're doing anything new or different to be able to create value for the organization, you are innovating. It doesn't have to be transformational. It can be really just an adjustment or a small adjacent thing that you're doing. But I think we s- what I see is in a lot of cases, you are leading with the technology rather than the business problem.

And so I do think we're seeing that [00:11:00] with AI as well 'cause I sit on a number of boards. And what are you looking for? You're looking for the AI strategy. That's not, it's not really about the AI strategy. It's what are the business problems that we're trying to solve? What are the processes that could be made more efficient?

Where are there places in the organization that we are spending more time than we should? And so those are really ripe to be able to bring in these new technology solution. And I s- I've seen examples of where they create an innovation team, and to your point, what happens is you end up that innovation team's almost treated like a virus in the organization.

So it doesn't have the right level of high-level support or sponsorship, and it is trying to be able to change these ingrained business processes, but it doesn't have the authority or influence to be able to do that. And it's almost like the rest of the organization finds it, sees it and kills it, and it ends up, not being successful.

And to be... Innovation stewardship is really thinking about what is the purpose, what is the problem that we're trying to solve, and then [00:12:00] let's look at the right technologies. And when you get into AI, again, I find AI is a bit like innovation. It's a really broad word, and it means a whole number of things, but you don't have to be way over in generative AI to be able to help solve problems, improve productivity.

And I think to your comment around people, how do you... we heard a very strong negative around AI's gonna come and take your jobs. If AI's gonna come and take your jobs, you're not very motivated to train the AI, right? And so now you're seeing much more of a human in the loop, augmented human, helping people be able to do more of what they like to do, automating the things that nobody enjoys anyway.

And so that becomes a much better narrative for humans and AI working together to be able to improve the outcomes for the organization. 

John Willis: Yeah. I think that, I, as, we discussed a little before I think a lot about systems thinking. And, one of the other very, not only... Totally, l- [00:13:00] 100% agree with you.

The leadership, like if we don't, if- this virus group is literally going off with, they've got the wink and the nod to do it, but nobody else really is clued in on the... But then there's also, even when the leadership says, "Hey, we're all gonna do AI," and then you got, team A which is dependent on team B, which is dependent on team C and D and on, and everywhere i-in between those are these little pockets of they're not really bought in, but they're passive aggra- so that becomes an integrated problem.

The other thing I think, and I don't know if you're seeing this in some of your sort of work with OCS or in general as an advisor, is that I think we're hyper-focused on the wrong metrics with AI right now. We're s- we're doing calibrations like oh, we were at 20% AI last year, we're at 40% AI this year.

Or, even to the extent where people are using token economics as leaderboards to say, "This is how much I- AI we're producing because we're doing this much code or we're doing, we're [00:14:00] using this much inference." Those are the wrong metrics. We've learned this from early days of code. In other words, it doesn't tell you the value, it just tells you how much.

And I was wondering if you had any thoughts there of seeing some patterns in your world. 

Kendra MacDonald: Yeah. So that's a great question. Actually, I just finished a team offsite, and so one of the things that we were talking about is we're investing in more AI tools, but we are not investing them just for the sake of investing in them.

We are actually looking to get a return. And so how do we measure return? And so how do you think about what is a thing that you are doing? And then if you are able to take that from four hours to 30 minutes, now you have saved three and a half hours, and we need to actually capture those. And then if you have that example, how do we replicate that more times across the organization so that you are, as with any business decision, we are making an investment to get a return.

It is not about how many tools we have or how [00:15:00] many AI-enabled things. If, like your phone, you give everybody a cell phone, they only use 5% of the power of the cell phone. My daughter would probably tell me I use less than that. In terms of how you're using the tools, then you're not actually maximizing the investments that you're making.

And so you just, you can turn AI into just a bigger, more powerful calculator if you're not really thoughtful about where you are inserting it in the business and how you're changing the business process to actually take advantage of of the tool. 

John Willis: Yeah. No, I think that, that I think the-- one of the bad assumptions that's happening with leadership is they think more AI is value.

In other words, if I have more AI, don't I by default get the, 15 hours or two weeks down to one hour? And maybe, not necessarily. Yeah. No I agree. 

Kendra MacDonald: And I think we're in that moment of a bit of disillusionment because you are not consistently seeing the outcomes, and so that will, slow things down I think in the next little while if people are not reframing the way they're having the conversation.

John Willis: That's right. [00:16:00] Yeah. And I always go back to learning. Are you learning? I use the J curve a lot as an example. I read all these failures of AI and I think, okay, are we in the bottom of the J curve right now? 

Where we're... and there's two ways to react when you're in the bottom of the J curve.

It's not working, maybe we should stop doing it, or we're incrementally learning, let's keep learning so we can accelerate out, right?

Kendra MacDonald: Absolutely. 

John Willis: And then I guess the sort of one question then is it seems like you talk with a lot of executives and leadership and that, we're all confused right now, right?

We're, there, there's 'Cause we'll get into governance and risk about AI here in a little bit. But but I think how do you address the, i, we've probably already answered it some, but I'm being redundant here. But, like, I've heard recently a CEO who said, "I think our organization is overwhelmed by AI and I don't think we're capable of understanding it as an organization, so therefore I think I have to bring in, a, one of the large advisor companies like Ernst & Young or Deloitte or whatever."

And I [00:17:00] think, like in my conversation was, have you even discovered your organization yet? And I think there's just they're in this confusing state where they've got a lot of branches to make decisions, and how do we help calibrate them to make better decisions and not knee-jerk decisions?

Kendra MacDonald: Yeah, I think if you don't know the problem that you're trying to solve, then any tool will help you solve it. And so I come from Deloitte by background, so obviously believe that the firms can be helpful, but you need to understand. So are they helping you to improve your financial processes? You think that there's a big opportunity in being able to optimize your supply chain or your materials management, or where do you think that...

and we've seen this, AI is the latest, but think about big data. And big data was going to be the thing that was going to make everything better. Data visualization was going to ma- be the thing that makes everything better. Any [00:18:00] tool, if you're not actually thinking about the business problem that you're trying to solve, then you are spending a bunch of time and money and creating a lot of frustration.

And so if you can anchor it in the problem you're trying to solve, and so now all of a sudden for those people, you are making it easier, you are making it better, you are saving them time. It also helps with the business case for people to buy in because they can actually see the value that it's going to bring to them as opposed to, wow, I need to get my head around this thing and I don't understand it.

So- 

John Willis: Yeah. No, I think it's organizational change 101. The I guess y- so your background is also in sort of risk governance, and I spend a lot of time working with organizations, particularly high consequence organizations, about the sort of the infrastructure risk when it comes to sort of everything.

But I've, my whole career has been around infrastructure risk, but for hospitals and banks and I think now there's a whole new layer of there's sort of two ways I think about AI, and that's not, that's what I think [00:19:00] about it's a sort of general categorization, is there's inference.

There's the ability to be assisted in knowledge, right? That's our chatbots, our sort of ask, get an answer, which can be o- very operational, right? And like on a ship, right? The the answer to what to do next could be catastrophic, right? ... And then there's this new layer of what's going on, what they call agentics where we're not only allowing the inference to give us knowledge to be able to make decisions about, or always human in the loop, but there's this idea that there are certain things we're gonna explore, and I think it's in its infancy, we're gonna do what they call human on the loop, which would, which like there's four things that can happen.

I think I can let an agentic process do the first three. But there's always gonna be risk because those first three decisions are going to be by definition inference-based Just like the original, even the assisted. So I think a lot about, as it seems like you work with a lot of AI [00:20:00] high consequence things, right?

Like ships and robotics and, even supply chain and all those things. And so what's the sort of the, I guess where I'm going is I think a lot about what a bank is going to have to do to show audit level of the things like you've got the wrong answer, but then you will get the wrong answer.

I guess that's the point, right? If you're using inference-based, there is no guarantee of deterministic answers. So if you're choosing to use AI in any sense where it's probabilistic, basically all AI is probabilistic, then there is gonna be a risk consequence. You have to deal with that. And then so one is, A how do you deal with it in high consequence?

Maybe don't do it. But if you are, and I think the economies are gonna force that and then, so how do you deal with sort of some form of containment? And then also how are you gonna have to explain when there's some type of consequence of the wrong decision from an audit perspective? That was a...

i'm a long-winded question. Sorry about that, but- 

Kendra MacDonald: No that's all [00:21:00] right. So maybe I'll probably tackle that in a few different ways. So I actually went back to school and did my master's in technology management a few years ago, and so one of the courses that I took was in AI, and the final question on the final was: Would you get in an AI-driven car, a self-driving car?

And it was predicated on the fact that who makes the decision, right? Now that the decisions around what happens in an accident or in a s- in that split second are all going to be preprogrammed, who makes that decision? Is every car gonna be programmed the same way? Would it be the way that you would actually react in the moment?

Are you actually comfortable giving over that decision not knowing what the car is going to do in that high-risk moment? And it was interesting 'cause my original answer would've been, "I'm absolutely going to get in a self-driving car." And then as I reflected on it, I thought, "You know what? I might wait a little bit longer and really try to understand how that information is going to flow through."

And so you're... same thing if you're [00:22:00] an autonomous ship, you get into that same type of thinking. And so we are in that moment that we're seeing with a number of our projects because we invest in these technology projects. What decision-making do you put on the edge in these sensors and platforms that are on the water?

What do you allow to start being an automated decision, and what still requires humans in the loop? And so my auditor background, what you're seeing more and more is you need that trust layer, so you have to understand what's happening in the box. And you need to-- We're going to see, I think, more and more trust and verification of that layer And you're seeing more and more accountability putting back onto the companies if the agentic or the chatbot or the whatever is coming up with an inappropriate answer.

You can't just say, "I didn't know," or, I never expected that." There is much more pressure, and so making sure that you understand what is going in, you understand what is happening and what is coming out. [00:23:00] And so I think you, you will see more and more tools that are doing the verification because that is the risk, is we hand over these decisions and we get these unexpected answers.

We don't wanna be giving over control of major assets on the ocean without being very confident in how those, And then you're going to see with agentic one agent that speaks to the next agent that speaks to the next agent. You're seeing that on social media where the agents are starting to talk to each other and there are some interesting things happening.

So roll that into a high-risk industry environment. You want to be very confident in how that's all going to play out. 

John Willis: Yeah. I think, it... Susie, I w- I was, like the one thing I know you know this is, to be crystal clear the point I'm trying to make is even though we think things like, oh, somebody driving their own car is a deterministic, it is not.

Humans make errors, right? And there is an ROI at some point where the crossover is that a m- a, an auto- autonomous vehicle could make less. I was in a Waymo just the other day and [00:24:00] literally I was in the bike lane pulling over 'cause I needed to see something, and I was literally pulling.

I was going slow, and I, what I was doing was illegal, but the this was something I saw or a sign or something like that, and that's even more illegal. But the Waymo figured that I... It was a two-lane, and it went around me and it came awfully close, right? And that was a weird edge case for the Waymo, right?

And- ... the guy in the backseat, just, flipped me off and was all mad at me. And- ... but that was a weird edge case that, you know, that any human could have made the same mistake. And then so I think again that, this idea of that, that autonomous decisions are dangerous or more dangerous than human decisions, I, I think that's gonna be a wash at some point.

But the other thing I think about that you s- that you pointed at is I think from an audit perspective, it's always going to have to be a legal entity or some type of entity. Air Canada was a great example, right? Air Canada with that sort of chatbot answer that they got, and they got sued.

Air Canada's argument [00:25:00] in the most simplest sense was blame the chatbot. 

And in the end, that you're not gonna win that argument, right? The ar- or you're not gonna win that lawsuit, or you're not gonna win a severe enforcement ac- action from a regulatory control body. You're going to have to show evidence that you were at least in all possibility in charge of the decision, and therefore then the decision went wrong, right?

And I think that's gonna be an important part of all, any sort of high con- any consequential answer or decision from an AI. 

Kendra MacDonald: Yeah. I think there, there is evidence or there's certainly studies being done to show that the back to the example of the car, 'cause I think it's one that's easy to get your head around, that y- you will reduce error.

Humans make lots of errors, and the cars make less errors. 

John Willis: Yeah. 

Kendra MacDonald: But they are also going to do the same thing every time, and so now you are programming in decision-making that you may not think of in terms of you're going to be in an accident. Does it choose you or does it [00:26:00] choose a pedestrian or it choose the pole?

What is it going to do? And so how do you make those decisions? And I think that due diligence argument in terms of being able to... So my background, again, I was, came back to Canada when Sarbanes-Oxley legislation was coming in, and so I spent a lot of time in the c- in the controls and processes.

And so you needed to be able to demonstrate that you had the controls and processes in place. So if something goes wrong, things will go wrong, but you had put the due diligence in place. You have your AI committee thinking through the various aspects, and so you can draw a line and say, "We did everything we could possibly do to be able to mitigate this risk."

John Willis: Yeah. No, I think that's, I always say that, like SOX is a good example, and I hadn't really used that, but is that, there's like your known knowns, right? These are things that you should know about, you should have and so when we find that something goes wrong, if you had a known, okay that became a known unknown, so therefore th- you your response sh- your, in fact, the finding is let's, here's it is.

This is not good. Don't let it [00:27:00] happen again. You got a time to address it. It should be a known at that point, right? And so there is- Sure ... this learning aspect to audit that, that, under the right conditions, not audit theater, which is a whole nother podcast. But the it is actually a learning, a learning sort of structure.

And I guess the, Oh you know what's interesting? So what are some of the interesting stuff that's going on with AI or autonomous capabilities in, in the marine world or anything related? That'd be interesting to hear. What sort of state of the art? What kind of really exciting things?

Kendra MacDonald: Yeah. So we have we've now supported over 150 projects, and now over 50% of them is some kind of AI application. And as an example, you've got satellite data being used at a port to be able to more efficiently park Or birth, birth the ships so that you have less movement and therefore you b- burn less fuel.

We've got similarly you're [00:28:00] using AI to be able to do net inspections in in aquaculture. You've got AI being used to be able to improve surveying for offshore wind farms. We've got increasingly called domain awareness. So how do we understand what's happening? So how do you get hyper-localized weather information?

Again, in shipping, if you can optimize even the ship's movement by a little bit, that actually has significant impact in terms of both cost and efficiency. So you're seeing it. W- we are a collaborative model, so you have more than one organization working together. And so one of the collaborations that works really well in our case is you've got a platform.

So you have an autonomous platform. It might be tethered or untethered, so is it connected or unconnected? It might be on the water, it might be under the water. And then it has sensors. So those sensors might be collecting acidity information. They might be collecting temperature information.

They might be looking at [00:29:00] levels of carbon genomics data. So increasingly you're able to cl- collect DNA in the water to be able to understand what is moving, what is the biodiversity. So in this movement of marine p- protected areas, you wanna make sure that they're being effective in protecting what they're meant to be protecting.

And then you have a customer that's looking for data. And so maybe the other variation of that is digital twins. 

So digital twins being a virtual replica of an asset. And for vessel maintenance, now you can be onshore and you can help with, because you can, virtually diagnose, and so you can do things onshore that would've required you to be on the ship.

So that creates efficiency in the maintenance process, but it also allows you to increasingly use predictive maintenance. So now I can estimate based on changes in behavior, how I swap out or deal with an asset before it actually breaks and the cost gets much greater. So I find it fascinating in terms of how many ways and the breadth of the various use [00:30:00] cases of AI across the ocean economy.

John Willis: I think the narrative is pretty awesome. I'm gonna still hold off to that last question, but the question I mentioned earlier, but I think your background of experience in all these technologies over the years and then applying them to a s- a space that is absolutely fascinating, 70% of the Earth, and then, seeing that sort of play out and even just what you just described sound...

i'm gonna take a real risk here right now because I don't even think the people listening to my podcast are probably tired of me. I'm writing, I wrote a book about the history of AI two years ago, and it's been out for a b- a little over a year and a half. And and the, But now I'm working on a very slow roll history of quantum computing.

And I did I've gone to a couple of large quantum conferences, and one thing I noticed is that the, in Canada, there, there were two very innovative quantum approaches in Canada. One was wildfires and forestry, and the other was the sort of the the power grids and the complexity of some of the areas in northern Canada [00:31:00] are just so spread out that it's a much harder problem than a municipal city.

So I, it seems to me everything you just talked about is, just somebody's waiting in the wings if they're not already doing something for quantum algorithms to address. Is there anything interesting going on there? 

Kendra MacDonald: Yeah. So we're s- we are-- So Canada has quite a bit of capability in quantum, and so we are starting to see...

So another use case is quan-quantum sensors. So where can quantum sensors have significant impact? It is actually in the ocean. And so as we're seeing an increase in defense spending and a lot of focus on dual use technologies, so quantum sensors has increased sensitivity, and so we are starting to see quantum sensors coming into the ocean economy.

So not as much as we've now seen with AI, but it is I am surprised as we talk about these technologies, that there's not more awareness of how many applications and opportunities there are across the ocean economy. So I think we're just scratching the sur- Yeah ... surface of how [00:32:00] quantum will start to, to come into various applications.

John Willis: Yeah, 'cause that's the key about quantum is it gives you the first time capability to solve these so intractable problems, the, where the the one I always think about is, if you can land planes consistently at a 30% degree angle, there's hundreds of millions of dollars to be saved by airlines, and that's an intractable math problem.

That's No, cool. That's really interesting. I guess the other thing you talked about that, that hit a chord with me is I don't know if it was your quote, but it was a quote that sort of came out of something in one of the articles that you were on a podcast or something with, but this idea of the governance slows innovation trope.

Yeah and that just, the fact that we still have to explain that to people in 2026 is just seems arcane, that would people do- still see governance as something that slows down innovation as opposed to a systems thinker or a complex adaptive thinking would be like, it actually speeds it [00:33:00] up.

Kendra MacDonald: Yes. So we, again, we had this team meeting and the go slow to go fast, right? So setting it up and one of the examples that, throughout my career, when you get into, you're always the risk people. So privacy, cybersecurity, and so you don't talk to those people until you get to the end of your systems implementation, and then they are the bad people that say you didn't consider this, and this."

And how much more effective is that solution, and how much more bulletproof is that solution if you're actually bringing in that conversation and the proper governance and controls and considerations at the front end? And I think we continue to see that. Same with legal and regulatory and, the more that you are able to build that in, the better that you are protected.

You're able to manage the various risks within the system. And so I think we still have that mindset that, if we bring these people in, it's just gonna slow everything down. Actually, if we get this right from the beginning, then we're able to accelerate because we have all the right people at the table.

And I think AI is exactly [00:34:00] the same way, is, the back end risk of something going wrong because you haven't properly put your steering committee in place, you haven't brought all the right people to the table, you haven't thought about the bias that's already inherent in your data. There are so many things that if you think about them up front, you are able to go faster in the back end. 

John Willis: Yeah. And that sort of leads me to one of the sort of other topics, which is, and, data. It's such an important part if we're gonna do AI, right? Like you talked about the inherent bias in the data but just all the things, just that how data can be tainted and, when we talk about risk, right?

One of the things that, that I encounter quite a bit is I cover IT risk in terms of people are reading sources of data. The adversaries are very clever right now in that they know how to taint data so that when it winds up in an inference, it actually can do nefarious things.

And so that there's a whole lot of sort of work going on is, like, how do you evaluate data at every stage of a pipeline [00:35:00] before it gets into a sort of an inference answer to a question, or even worse, a decision for an agent to talk to an agent to tell it to do something? But I noticed you talk a lot about data as a trust and data pipeline, and so I, the same issues arise, right?

What... if there's an adversary that wants to somehow taint data that somehow winds up in a supply chain of an autonomous ship, right? So I- that anything that you are th-thinking about, talking about, or people are doing now would be of interest that as a learning opportunity for other high consequence.

Kendra MacDonald: Yeah. So I... So a couple of things. So big- couple of big challenges in the ocean right now and ocean data. One is a lack of data. And so if you think about how much data you need to do training models, we had one of our companies that actually- Built their algorithm. They had to figure out how to use less data to train.

And so they've actually been able to move from fisheries detection to [00:36:00] object detection, and that has allowed them to broaden out their application because they were dealing with this lack of data. Interoperability is another big challenge. Can you go a little 

John Willis: deeper on that, the fisher- fishery data versus object data?

That, that sounds fascinating. Yeah. 

Kendra MacDonald: Yeah. So the company is OnDeck, so it was OnDeck Fisheries, and it actually started as a $10,000 prize to a group of students at the university back in 2022, and their idea was, so in fisheries you have fish- fish inspectors, and so they get on ships and they look at what you're catching and make sure that you're in compliance.

A very heavy human- okay ... lots of subjectivity to error. And so they thought, "We can bring computer vision to this problem, and so we can then use computer vision to be able to look at the fish- 

John Willis: Okay ... 

Kendra MacDonald: and be able to count it in a much more automated way and count many more fish and truly ensure compliance."

And so as they were developing the training data they realized that it was a real struggle. And so this is probably where your [00:37:00] technical capability is higher than mine. But essentially what they then did is they said, "We need to reverse train." And they're I think NeurIPS, N-E-U-R-I-P-S, they had a paper that was published there in terms of how they actually were able to say, "How do we get the model to say, 'This is an object of a certain length with a certain shape.

Oh, we think it's a tuna,' or back into it?" Okay. And so that allowed them to broaden out the application of that. And so now they're working with the minister Ministry of Defense, for example, in Singapore and doing object detection more broadly, and they were in Y Combinator as well, And they've dropped fisheries, so now they're OnDeck AI.

But that just seeing how they tackled this this challenge of data. I probably lost now my train of thought on the original question as much. 

John Willis: Yeah. So the one was the fisheries object, but then you were listing out some other ways of some interesting ideas that talk about data and data trust and building.

Kendra MacDonald: Yeah. So we've actually looked at how... So one of the [00:38:00] challenges- So I think we are still struggling with thinking of all of these assets on the ocean as targets in terms of we have a project, for example, is y- you're putting wearables on your mariners. Now you have created a new surface vector for the ship, for example.

We've got all these sensors, we've got all this technology, and now as you're adding more real-time communication, again you're creating new risks in terms of the collection of data. The other thing is how do you... and this is a conversation in many of the sectors, is you want to improve safety, right?

And so where is your safety data? Is it a competitive advantage or do we-- can we combine all of that safety data so we make the whole industry safer, and so that is the type of data that we can share. And so we're in this constant debate or discussion around where is it value add to share. Where is your data truly [00:39:00] competitive, and where would there be much more value to you and the industry and reputation, et cetera, if you're able to share the data.

One of the other courses that I took was in marine law, and I find the whole area of shipping really fascinating because you've got hundreds of years of law- 

... 

Kendra MacDonald: That is predicated on a ship owner that is on land, a captain that is at sea with a crew full of workers on the ship, and there's no communication.

Like that-- the law is based on that. And so now you're fundamentally changing that because you've got constant communication, so decision-making can flow between the captain and the ship owner. And then in some cases you've got no one on the ship, so now you've got an operating center that is controlling a ship that is in one co-- one country.

You've got a flag from that-- on the ship that's another country. And so what does that do in terms of- What do you have access to? Then you've got whatever port it's in and local police, and then how does that all work together? And so you're seeing this tremendous ship as [00:40:00] we-- shift, not ship. Yeah.

Tremendous shift as we move into economy that is trying to unravel or overlay on its hundreds of years of no data and no ongoing flows of communication into a very traditional industry, and that's just one example. 

John Willis: Yeah, no, those are great ex- I think about, I remember my my, my oldest son when he was about four or five, he asked me, "Why does, why, how come cars can't fly?"

And he's the, the... How do you answer that sent to a five-year-old? They could if there wasn't all this other stuff around how would we regulate it, what, how would we, all that kind of stuff. I... The one thing that you you hit a nerve with me, which is I go back to my Dr.

Deming and, one of the things he's famous for is going over to Japan after World War II and helped create an atmosphere of some people say, helped create the miracle in Japan that we all saw in the '80s. You, as, as Ja- Japan's invasion of everything, cars, televisions.

One of the things [00:41:00] he did over there was, laid down the gauntlet that now is not the time to compete. We have to work collectively. So he literally had to break... and it was systems thinking. It was like, do you want to be a world economic power? And the, in fact, his question was if you want that to happen in five years, and it actually did happen in five years then this is how you're going to have to do it.

So when I think about, the question, it sounded like you, that people are still reluctant to share things as simple as safety data. It just seems, at first it's the total anti-pattern to anything Deming would try to teach, but so is a lot of other things. But more importantly is there resistance still to sharing safety data and wh- why and how do you break that?

Kendra MacDonald: Yeah, so I picked on safety. I think safety is an area where you are seeing more and more- Okay. Okay ... 

sharing data. I think where we do see it is when you get into broader understanding of the ocean, and you've got these data sets that are [00:42:00] contracted not to be shared, so- Okay ... thinking about what clauses we're putting in those contracts, for example.

I think your comment around competing versus collaborating, so one of the, the model in Canada is really how do we get people working together and thinking about how do we make the whole country stronger, which is similar to the example that you just shared. And then I think with, if you look at the ocean economy overall life underwater, which is one of the 17 sustainable development goals, is the most under-invested.

All right. So you're in a moment in time where you've got the ocean under significant pressure. It touches so many, communities all around the world, and then you are not necessarily optimizing or working together. And so how do you encourage not only that collaboration across the country, but where we can, that collaboration all around the world, so we're able to understand things faster and make better decisions for our limited resources?

John Willis: Brilliant. Yeah. I was thinking about what, how do we wrap this up, and there's sort of two ways. [00:43:00] One one is more of a meta sort of question. If, with all your experience of, working with risk and organization and new technologies over the years and and now seeing this supercluster and observing that, 'cause I always think those are good Petri dishes, right?

These of a, that sounds like a brilliant Petri dish for applying a lot of ideas and that's what fascinates me. Then if somebody was to bring... A friend said, "Hey, you really need to talk to my CEO of a bank or healthcare or," w- is, and I, this is a terrible question as I think about it, but what would what would be three things that would come off the top of your head of what would how would you try to help navigate them with maybe three questions or three points of advice or 

Kendra MacDonald: Wow.

That, that is a big question. A terrible 

John Willis: question, yeah. No, I'm joking. 

Kendra MacDonald: I would start with have you thought about the ocean? Oh. So that is actually- Wow. I know, yeah ... a lot of cases. I was just in an [00:44:00] audience on Tuesday and said- Awesome ... here we are, Canada, and three of our borders are water.

And so the ocean is critical, and the Navy is critical to the conversation in defense, and that was a, an aha moment. That's amazing. And so you would think Canada has the longest coastline in the world, how would ocean not be part of a defense conversation? But I think the same when you get into climate and you get into impact investing, you get into clean tech solutions.

I think people don't necessarily think about the ocean. I think the second piece is even a sub-layer of that is no matter where you are in the country, 'cause I think sometimes we treat the ocean as a coastal problem- 

... 

Kendra MacDonald: You are impacted. So you are impacted by the changes in weather. You are impacted by supply chain.

We are seeing this playing out internationally. You are impacted if your internet goes down because our subsea cables are at risk. And really understanding what, how those sectors operate and how you are interconnected in terms of the ocean. And [00:45:00] then I think the third is really how can you get involved?

And so there is a role for, because we touch so many sectors, because you've got so many technologies playing out, every skill, and we, we need the investors. We need the talent in terms of the technical competencies. We need the traditional scientific knowledge that we have and need in the ocean because there's so many scientific questions.

We don't want to make a decision over here that just causes a- another problem over there. And so how do you bring that skill into this problem that we are facing? Because it is a tangly, big, complicated problem. We used to think that the ocean, talk about a system, we used to think the ocean was too big to fail.

And we have now proved that is not the case. And so we are in a moment in time where we need everyone to be part of that, not only for a stronger economy, but also for a healthier planet. 

John Willis: Yeah. No, that, that's great. Yeah. No, I think that all fits. I guess I'll, end with do you have any questions for me?

Kendra MacDonald: I'd be very interested [00:46:00] from your lens as you're thinking through AI and quantum, are there any lessons from that I should be thinking about in the ocean economy? 

John Willis: Yeah, no, I think quantum is I have quantum on the brain right now. And but and I'm just absolutely consumed by it.

And but I do think that it's that everything you mentioned on the things like, being able To do these sort of equations. Think about like the equations that even with the most advanced CPU, you know, graphic card, Nvidia cards, all that, these and, and little slices of answers that you can get are, best case hour of computation, maybe more than that.

Imagine y- you can answer those questions in seconds. Imagine you can answer those questions in secon- and we're still a ways off, but it's hit, it... the thing I tell CIOs is I like, "Don't drop everything." All right, let me step back. The, if you had a chris- I tell people that what we have today is generative AI.

You could argue the real [00:47:00] sort of spark plug of that was basically 2015, I think, where Google basically wrote a paper and it basically said "I think we can do neural networks a certain way," right? And it was called Attention Is All You Need. It's just technical jargon. But that started a ramp of what became OpenAI, which became GPT-2, 3, and, 4, 4 and now 5.

Anthropic, those people the person who started Anthropic came from OpenAI, all came from Google. A question I ask a CI- a CIO or really a CIO and even CEOs, and the same question I'd ask you is, if you knew now, if you knew now what you could've known in 2017, what would you have done different in a preparation?

You probably would've hired more AI talent in 2018, '19. You would've in, and not just put them on the problems that at the time we thought AI was gonna solve. You'd put it on all the business problems. You'd literally say, I'm gonna hire [00:48:00] quantum talent out of the top universities, and I'm not gonna throw them on the sort of the obvious, quantum proofing and RSA encryption proofing or whatever.

I'm gonna throw them on the, in your case, Marine. I wanna sp- shove them all over the place, and I want them to then start thinking about what they know about quantum and the intractable problems and how this thing could play out in some near future," which could be by most quantum experts, which mostly are ones who have quantum products, so there's a double-edged sword there.

But they're saying about five years out that we might be able to have like quantum supremacy. And even if they're off by five years, that's 10 years where you could solve problems that computationally would take, some cases billions of years can be done in days, right? And ones that could be done in a thousand years could be done in minutes.

And Google, even though they're all Petri dish and they're very isolated experiments they just they just they just simulate a dr- [00:49:00] a sort of protein bonding of a 1,200 atom molecule through this hybrid solution. So the it's coming. And so all those things that you listed out as optimization.

Yeah, and again, that's the point. Optimization is going to be key for quantum. We're gonna be able to optimize in ways that were just unthought of. And so I would, and what I would tell, what I tell CIOs and cer- some CEOs is, "Don't trap everything, but carve out a small percentage of your talent and roadmap to understand this thing that's going to happen.

And when it happens, you'll be more prepared for people who just wait till it to happen and then start ca- playing catch up," which almost all of us have been doing in AI. 

Kendra MacDonald: So I think a really interesting space we haven't really talked about, but there's so much biodiversity still to be discovered in the ocean, and increasingly you're looking to the ocean for beauty and pharmaceutical inputs because as we get our super [00:50:00] bugs, we need to find these solutions that are in the ocean.

And so you'll be able to do that at a different speed, I think, with quantum. 

John Willis: Yeah. And problems that just are not really, they're not computationally obvious or even tractable that you can now discover, things that, you know... in the this, some of the seminars I saw about the wild life and fires are being able to calibrate or computate p- prediction of where a fire is going- 

At scale, right? Those are things that you can run back on models, you can run them for hours, and you could say at that time the fire's already gone in this direction." But what if you could do that sorta, in a perfect world, which again, we're far off from that, instantaneously, and then yet be able to discover things that we just, you know, you- humans couldn't comprehend.

I'll stop here in a minute, but the whole point of this is humans, even in all our sort of math and computation and everything we've been doing with computers for 75 years, basically in anger, [00:51:00] is all based on our notion of the physical world. So we have a container of how we physically think about things.

What happens with quantum computing, it enables a container we're not aware of how to solve problems. Even to the point where you could use quantum to try to solve a problem where the output is still not understandable by a human, but you take those, that output to go into AI to then identify the patterns that a translation for us, right?

And so the, anything that you can think about as a possibility with quantum is a possibility 

Kendra MacDonald: Yeah. Tied right to your fire example is weather, right? So as we get more extreme weather, the faster that we can do that, then we can predict when- Oh, that... 

John Willis: Yeah, that's right. Yeah. Just and even the pragmatic approach of being able to, qu- quickly predict where the fire is going, how many lives get saved, how many houses get saved, right?

Yeah, good. And then I, I will put links [00:52:00] to your organization. Is anything anywhere special you'd want people if they they get uber excited about this conversation again, we'll put all your bio links and all that in the... But any place particularly? Or do you want to... You gotta beware, I do podcasts with people where they're kinda like, "I like John that we did that podcast, but boy, now I get calls from a lot of different people."

'Cause they're inquisitive. With my group, like there's not a ton of them, but boy, when they get excited they'll cl- they'll literally clamp onto you and wanna pick your brain. 

Kendra MacDonald: Yeah. So ocean, oceansupercluster.ca is where you can get lots of information on some of the projects. I just scratched the surface of our projects.

If you're looking for me, easiest to find me is on LinkedIn. I also tinker a little bit with Substack and just what are the investment opportunities and why should we care about these different areas of the ocean. So that's something else that you could check out as well. It's Saltwater Signals, and you can find that on Substack.

So there's a few different angles to the conversation. 

John Willis: Brilliant. I, and I [00:53:00] hope you had fun. I definitely had fun. It was a lot of learning for me, so that was pretty cool,

Kendra MacDonald: yeah. Thanks so much. 

John Willis: All right. Thank you.