The AI Gold Rush | Steven Harrison on How Data and Automation are Altering Business

Hello everybody, welcome back to another episode of Mastering Tech Growth. Today we have Stephen Harrison from InfraTech Digital, and we will be talking about data automation and how AI impacts our economy. Welcome, Stephen.

Yeah, thanks, Mike. I'm glad to be here.

Of course. Let us know who Steven is. How did you end up in tech industry?

Sure, sure. So I'm one of those, I'm one of those techies that, as they would quote a famous character on a famous sci-fi show, that I might be a captain by rank, but I've only ever wanted to be an engineer. I started my career in tech. I've always worked in tech. I've never worked outside of the industry. And I've always just had a real fascination for how digital technology works. You know, it's again quite unusual that I was not attracted into the writing apps and software and things like that. I'm more of the classic techie that likes to pull things apart and find out how they work.

How did you end up— so Infratech Digital being a company you co-founded implies by the name that is heavy on infrastructure. Am I right to assume that?

Yeah, that's very correct. So one of the things that I've always worked around is the various bits that make our digital world possible. And it's almost been forgotten nowadays that, you know, you pick up a smartphone or a PC or you log into an app, and what you see is that lovely user experience, but you forget that somewhere deep down underneath there's actual equipment in a building somewhere that makes all that stuff possible. And without that, you don't have this lovely digital world that we live in. And so my passion being in that area of, as I said, pulling things apart— how do they really work? What's underneath the experience that the users get? That's what really kind of drove my career, was always looking at how to design that better and how to get more resilience and more security and more performance out of the underpinnings that make everything else possible. And a few years ago, realized that there was a real gap in that knowledge space. If you look at our educational system today, We don't teach how to build the things that make digital possible. We teach how to consume digital when it's already been built. So it begs the question of, well, who's going to build the next generation?

That's such a nice summary, Stephen. I love that. When you were talking about UX, people open the apps, you know, they see the user experience, they consume the technology. And you, you know, you talked a little bit about what happens under the hood. And I felt like that appreciation is just lost because I'm coming from this backend industry as well, where, you know, we're writing everything in the backend and we never get so much claps, you know, compared to like product people who build their UXs, right? It's always like, we build that stuff. It doesn't run without us. Such a nice, um, well, and it's true what they say, right?

That the real infrastructure people. Are never noticed until it fails.

Exactly.

And that is a terrible thing. It's true of all types of infrastructure, isn't it? We don't really appreciate our road network until it's full of potholes, and then we go, what happened? This is terrible. Well, yeah, right. That's because we haven't been building roads for the last 20 years, so now they're all falling apart because they do that. We actually have to go and rebuild them once in a while. It's an important aspect of infrastructure. And I think we're starting to see the first few potholes appear in our digital world. You look at the real increase in the number of breaches and cyber attacks and things happening at a nation-state level now in the digital world. And you think, those are the potholes that happen right before the road washes out from under us? Are we paying attention as a society? Are we going to— get back to training people how to design and build roads, or are we just going to keep letting things deteriorate and put a nice top dressing on top and update the paint marking out the cycle lanes?

That's a very nice analogy, Stephen. So that comes from old school economy, isn't it? If we talk about Adam Smith's The Wealth of Nations, he talked about what actually builds the nations, right? So we have Do you want to introduce this theory? Because I know you love it.

Yeah, no, absolutely. So I, again, strange thing that I'm a techie, but I also kind of like finance. So, you know, my hobby has been studying economics and finance and how value is created. And while there's a lot of different economic theories, Adam Smith's theory I think still holds broadly true today. They said if you want to talk about creating value, if you want to create wealth, you're really employing 3 different things together at the same time to a goal. So part number 1 of that is you have to employ some labor. There has to be somebody doing some work of value. That labor is going to require some resources. So that's going to be your raw materials, but also your building, your real estate, you know, a place to do that work. And then you're going to need some sort of capital. You're going to need money in order to pay for the resources and pay for the labor. Bring those three things together, and if you've done a good job, you've now generated value because you'll receive back more than those three ingredients put in.

So does that still hold in today's tech economy?

Well, and that's something that I've been thinking about a lot in recent years, and 2024 especially, we're getting a lot more discussions around AI. And it made me think that I think we actually need to update Adam Smith's little model there. Hmm. Because you know what, here's the thing. Is an AI contributing labor? Because it's not really labor, is it? That AI isn't limited by working a certain number of hours per day, or by doing one task at a time. If I need more output in the labor sense, I have to go hire more people, don't I? But if I need more output from an AI, I can just scale it. I don't need another AI. The one I have can simply produce more. So there's one new dynamic there when you talk about AI and any type of automation technology that says that's not really a resource or a labor. That's something new that wasn't around when Adam Smith coined his theory. And then you've got the second aspect to that that I think we're only just starting to discover now, and that's the concept of data itself being valuable, being an input. Because if you go back into history, obviously, you know, you would think data was just knowledge. People often think data, knowledge, it's the same thing. But it isn't really, because if that data was simply knowledge, that a casual observer could show up, look at what you were doing, and maybe, you know, back in Adam Smith's day, you were stitching a piece of clothes to make a new shirt, and they could go, oh, that's a good idea, go back to their country, hire some labor, and start making shirts. Easy. But data isn't really like that. I can't be a casual observer of data and then go away and actually monetize it because I wouldn't know how. I would need the actual data itself in order to monetize it, not just the observation of the data. And you see that already with, you know, large social networks and things like this. Where's the value in Meta, in Google, in these organizations? I can't simply copy what they're doing, can I? Right? I could launch a brand new website that looked exactly like Facebook. I still wouldn't be Facebook. Even with all of the capital, all the resources, and all the labor in the world, I couldn't be Facebook because only Facebook has Facebook's data. Hmm. So we almost go from a place where there's 3 ingredients to our economy to a place where there's 5 now. And that the wealth of our nation is going to be determined by, yes, still our labor, Are we effectively employing labor? We still need resources of a sort. We're still going to need energy, space, buildings. We're still going to need capital. But we're also now going to need data, which is unique and powerful to our own economy. And we're going to need AI that is again unique and our own in order to leverage competitive advantage in the world scene.

Can we look at AI and just say it's labor, but it's just much cheaper?

I'm not entirely sure we can.

Okay.

I think today, absolutely. But I think when people do that, they're missing a bit of the what could AI do for us? Because one of the things that has been sort of the fallacy of any economic transformation, and we saw this back in Adam Smith's day with the textile industry in the UK. At a time, the UK was the world leader in textiles. If you wanted silks, if you wanted woolen garments, any of these things, you came to a British supplier because we produce the best product for the best value. You look at today's market and you think, why don't we have a textile industry in the UK anymore? Where did it go? And I think a lot of what happened is that when electrification came in and was displacing the original Industrial Revolution, the steam power, water power, things of that nature. What they did is they said, oh, I can use that electricity to run my cotton mill or my silk mill instead of coal or instead of steam and continue running the mill that I had yesterday with a new source of power. What other parts of the world did is say Hang on, that electricity lets me design an entirely different type of mill, and that's transformative as opposed to a simple evolution. A lot of people are making the same mistake with AI today, is they're saying, well, today I employ this call center worker to answer customer queries, and so tomorrow I'm going to employ this AI to answer customer queries. What they're not asking is, can I transform the customer experience to eliminate the query entirely?

And you're saying AI will enable this eventually?

Yes.

And just business. So effectively, businesses who will approach AI in a creative ways, they will make another step forward and another step forward, which doesn't make AI just about automation. It actually will enable a different business vertical. So just a different way of solving problems like Exactly. Okay.

Exactly. And we're seeing, we're seeing little bits of that in other parts of the economy already. And, you know, good examples being in additive manufacturing where they're now realizing that the design process for making a new part or a new component for a car or a spaceship in the case of SpaceX, that they can actually work with an entirely new design process now because of additive manufacturing. So instead of saying, well, that 3D printing means that I can save time and money making this traditionally designed part, they've turned that around and said that 3D printing means I can design a different part. And that's transformative. It's the same technology, but I've used my imagination to step outside of the, you know, as they say, the 8 most expensive words in business. This is how we've always done this. I love that. I can use that technology to completely step outside of that dynamic and say, is there a different way? And that's where AI is going to transcend just being another form of labor, is it's going to give us insight and capabilities in our economy that 5 years ago we couldn't have even imagined. Now is the time for that imagination and to create that real differentiation. The businesses that do are going to excel. The economies that do are going to excel. The ones that don't are going to go the way of our textile industry in the UK and fade away into history.

Yeah, they will just— because it's moving so rapidly. It's either you're part of the race or you will just not be able to catch up.

That's exactly it. You know, at the point that things have moved on, it genuinely will be too late. There won't be a catching up.

Exactly. The distance will be too far. So before we dig into AI, Stephen, Data, right? So you went and you went like, this is the fifth one, right? So, so we have labor, resources, capital, data, and automation. So when you look at data as a resource, just another way of helping you get things done. No, no.

And here's, here's again where I think data is different enough for us to give it a unique type of focus. When we're planning our businesses, when we're planning our economies. Because the big thing with resources, and the reason it's one of the core three elements in Adam Smith's model, was that resource was finite. So if I had this many square meters of land to build my factory, that is all I had. I could build my factory taller, but I absolutely could not build it any wider. Now what's interesting with data is that data is probably for the first time ever an infinite resource, right? I can't take a raw material like cotton or gold or steel and go copy paste and suddenly have twice as much of it, but I can with data, can't I?

Yes, you can.

And so that then suddenly gives me new opportunities and new ways to examine what I'm doing with my business, because the exact same piece of data manipulated slightly, becomes a different product. Manipulate that data again, now it's a third product, a fourth product, and I can very rapidly start to explore other avenues of business. And I think, well, you know, without being overly cynical, that that's one of the pushbacks that people have had with the big data-heavy businesses of today, the Googles and the Metas, is they're doing exactly that, aren't they? They're looking at the data they already have and going, oh, there's unrealized value in that data. I can make a copy of it, change a few things, and sell it again. And as individuals, we're going, oh, hang on a minute, we don't like that because you're now monetizing my data. But we've got some deeper questions to answer, don't we? Because questions like, well, is it your data? When you signed that user agreement, when you agreed to join that application or that service, did you not agree that they were going to collect that data? How can they put a box around how they're going to use that data in the future? These are very difficult questions to answer. And what we've seen with, for example, very simply, the European GDPR rules that we passed The result of that is a lot of international businesses simply block access to Europeans now. And so we've created a multi-speed internet, if you will, where if you live in Europe, lots of different websites and applications say, "Not for you." But other parts of the world can use those sources of data, can use those applications, can use those automations to accelerate their economy and to grow their value. And we've effectively removed ourself from that opportunity because of these concerns over who owns that data and how is it going to be used. That's not to say the GDPR isn't wrong, but I think that we need to do some further thinking about those topics because it's going to have more and more knock-on impact as more of the global economy starts to appreciate data as being really different than a traditional finite resource. As something that we can exploit over and over and over again without having to go find more of it. Hmm.

So very good, very good explanation, Stephen. While you were talking, I was thinking about data ownership and you touched it a little bit. Does it simply mean because you always tick a box, you always go like, I'm okay, I'm okay.

Everyone does.

Exactly. And no one really reads the terms. I mean, honestly, how many people actually read through the terms before going like, yes, I agree to actually use the service either online online or offline. Everybody just ticks the box. At the end of the day, if it's free, you are the product. So we know that. And does it mean that being online is effectively selling what you do, what you consume, just by default? Like, if you come online, the moment you enter internet, you gonna be somehow monetized. Things you do, things you read, things you buy, anything you express as me, it's Someone will take it and use it.

There's certainly an argument to say that that is the case. I think that the bigger philosophical questions of is it right, is it wrong are things that, you know, we need to have a serious discourse about in society. I myself, when I engage with online services, I engage in the full knowledge that everything I do is going to be monitored, tracked, and monetized. And so if I'm concerned about that, I will either not engage or I'll use some sort of an anonymizer, right? So a VPN service or something of that nature. And that's my choice, right? I have chosen to engage, and I have chosen to engage either as me, as my natural self, or as an anonymous version of me opening an in-private browser window or using a VPN or doing something of that nature. To conceal my identity and my activities. Um, that's something that I think is, in a certain extent, a bit of learned behavior that we need to express out into society. You know, we would have had similar things going back, you know, tens of thousands of years ago, of make sure you put clothes on when you leave your house because otherwise people will see you naked. You might be okay with that, they might not be. And we had to come to that societal agreement that said, even if you're fine being seen naked, your neighbor doesn't want to look at that. Put some clothes on. And so we've all adopted a convention that works for the mass. And sure, it's inconvenient for certain members of society who would much rather walk around naked, but they've accepted that they're in the minority, haven't they? What we have today, I think, is a lot of times we have a minority in the online community making a lot of noise and driving legislation in a direction that maybe the majority isn't that worried about. Maybe the majority would say, you know what, I'd rather just have less advertisements. I'd rather have less things I have to click through because I'm not worried about it. And maybe that very noisy minority we need to have a conversation with and say, 'Well, you don't like this, but you can choose to either not engage or engage differently,' and give people back some of that individual freedom of choice. The alternative is we end up with a very, very controlled internet. You know, we almost go down the direction of a planned economy, and that's going to make things difficult, isn't it? Because inevitably it will limit innovation in some way, shape, or form.

Yeah, it's not going to be a free market. So while you were saying this, um, I remember a couple of projects, uh, um, so I participated with. So it was a social media platform which effectively the whole unique sales proposition was that we don't monetize your data, but you will be paying for our service. So effectively we give you all what you get with Facebook or Twitter or any other social media. The feature-wise, it's, it's very similar. But you will be paying a monthly fee. As a result, you know, we're not going to be effectively selling your data to third parties or utilizing to, you know, show you ads or whatnot. You're just going to be paying for a service which all the features are for you so you, so you can use the service. No running ads, no nothing. And, um, always raise this question, you know, how, how much market there is, meaning how many people would actually be willing to pay a monthly fee instead of, you know, going using for free by being monetized as a data point. I don't have an answer. I'm just giving context. So I don't know how the project went, but I know that there was a competing at least like 7 projects. One of them were quite big and there is a segment of people who actually be willing to pay, but it is leaning towards that matured, aged society piece or people group who actually are aware of the, you know, data monetization and they, you know, they can actually pay the monthly fees and they go like, you know what, I just don't want to be in that public profile. I'd rather pay and remain in this more private profile. Would you then say that From your perspective, Stephen, there is an element of I'm young, I'm coming on the internet, maybe I have no idea. So we effectively, if we're not giving the knowledge or education about what will happen the moment you go online and how the data will be gathered about you, is something we should do as a society?

Just very much so. Yeah, I think, and coming back to where I started in the beginning, my my passion being infrastructure, that I think there's a huge lacking in education for how the digital world actually works. Um, I, I remember one of my favorite things years ago when people first started the concept of the cloud. Now, of course, the cloud, what it actually is underneath, is not new. We've been doing it since the 1970s. We just had a different name for it. But marketing promoted this as this new thing. And so now it was in the consumer space, it was on people's tongues, and they were saying, you know, well, yeah, I've got my pictures in the cloud. And I would say, right, so where is that? And they would go, what do you mean? It's in the cloud. But it's not magic. Your photos are stored somewhere on something somewhere in the world. Where are they? And you can see the, the moment where the kind of the first sparks of light form, and they go, oh my, I've not actually thought about the question, let alone asked it. And I said, if this was your physical photo album, would you not want to know where it was? And they go, of course, like, I don't leave my physical items lying anywhere, right? Well, why are your digital ones being regarded differently? Surely you want to know. Surely you want some say over that. When you bring that question from the individual up into large companies and governments and state actors, it becomes very, very critical. But if the people running these projects, the people that we employ in the public sector, in the large organizations, don't have this foundational knowledge, they never even ask the question, let alone look for a suitable answer, which means that you're allowing your service provider, whoever that might be, to make those decisions for you. And those decisions have profound impacts. So we almost need to start teaching some of this stuff as we would basic literacy. You think about all the other things as a society that that even a generation ago were not considered essential knowledge. Perfect example, reading and writing. In the grand scheme of human history, it's relatively recent that we thought of that as you must. That was always considered something of, well, if you're interested, but if you're not, don't worry about it. And we realize that no, as a society, that is a severe disability to somebody if they cannot read and write. That person is now in danger of exploitation, of, you know, economic slavery, all of these other things. So we made sure to teach them reading and writing. We're at a point now we need to do the same with digital skills. We need to be teaching that from preschool so that our young people understand that if you don't know how this stuff works at a high level, you are at risk of being exploited. You are at risk of being taken advantage of, of being deprived of something that's of value to you, and you won't know because you lack the skills to understand when somebody is taking advantage. If we all had more of those foundational skills, just as we've done with reading and writing, it would, I think, alleviate a lot of the fears that people have going into things, you know. And that's where people were previously terrified of contracts and lawyers, because if I can't read the contract but I'm being told to sign it, Well, of course that's scary. I have to trust that the person giving me that contract is trustworthy. So we taught that person to read, to take responsibility for themselves. That fear went away. We might not love lawyers, but we're not nearly as afraid of them as we used to be. Hmm. We're now seeing some of those same things in digital, that there's a lot of people in society, and as you said, some of the older people are suffering it the worst. Because they've not grown up around these things and they're afraid. Well, why are they afraid? Because they lack knowledge, not because the thing itself is scary.

By default, the lack of knowledge is scary. So if you don't know, that's where the fear is coming from. The moment you know, that would alleviate the fear. And exploitation, at least, I mean, I remember when I entered the internet, I mean, the viruses, you know, all the clickbait ads leading to dodgy websites asking to leave your data and credit cards. I never got exposed to it, but I always got attacked by it. But many, many stories around me, which, you know, a grandma, a grandpa leaving details, you know, PayPals in the beginning of PayPal, you know, getting hacked, people losing money. That's effectively from the lack of knowledge of how to actually behave and where is the danger on internet. And yeah, it's a good show.

And, and AI is absolutely going to amplify that because what's happening right now, people didn't understand digital to begin with. We're now adding AI on top of digital and we've got movies like The Terminator out there. So this lack of knowledge is just amping people's fear up to a whole other level.

Yep.

If we focus on that knowledge, we can get rid of the fear and people can start to see the opportunity.

Yep, exactly. And I heard that AI will take our jobs as probably the main one, you know, AI and Terminator future visions, you know, as a second one. And then you go like, it's just ML, right? Right. So before we dig into AI itself, Stephen, Last question about data. So if we look at the data as an actual wealth, right, it's a thing which you can actually monetize, and you rightfully pointed out that you can repurpose it to find different ways to generate revenue. If we look at data as wealth, do you think there's a bubble around it and it's actually going to burst eventually and become a lot less profitable? Or do you think it's a, as you mentioned, infinite resource which which creative people will be able to just mine forever?

Um, I think it's probably more of the latter. And the reason I say that is that the data is most valuable when the data is most rare. That's true of any resource, isn't it? Right? I value platinum more than I value iron because there's a lot less of it, not because it's less useful. Arguably, I encounter iron a lot more often in my daily life than I do platinum. But the rarity increases its value. Because organizations and individuals today are not being as careful as they perhaps should be with their data, there's a lot more of it out there. As we evolve as a society and we start to be a little bit more cautious about who can have that data, who's allowed to sell it, how many copies can they make, which country can they store it in, What are they allowed to do with it? What we're going to do, by virtue of this is how economies work, is we're going to increase the value of that data. So if you think about where we are today, I kind of liken it to the early days of the oil and gas industry. Oil wasn't terribly valuable when we first started using it because we didn't have a lot of use for it, but the people that saw that This new source of energy is going to transform the world. And they went out and what did they do? They laid claims to massive areas of the planet and said, that's all my oil. I don't even know if there's oil there, but if there is, it's mine. And they started to stake out those claims. And we saw over the 20th century a massive shift in the wealth of nations, the rise of the wealth of, you know, the OPEC states in the Middle East and places like this. Purely on the recognition that that resource might have value, that it would be finite, that it would be controlled at some point. And if I stake a claim to it early, I can grab more of it while it's undervalued, and I can then use that as an infinite money printer. Now, that one— of course, oil is a finite resource, so it's not an infinite money printer. Some point your oil field runs dry. Your data will never run dry. So the organizations that do that work best today, I think they're going to be positioned for an incredibly lucrative 100 years, because as data becomes more controlled and more constrained, the value of the people that do have the data is just going to go up exponentially.

Okay, so if you would be an investing person, You would just buy stocks for companies who are exposed to big data right now.

Yes. The companies that are consuming and processing and creating data, they're going to be worth more and more and more money as time goes on.

Okay.

That would certainly be my prediction.

Not a financial advice. Just throwing out there. But I think you're absolutely right. I think you're absolutely right. All right. So AI. In the very beginning of conversation, you connected AI with automation. Do you think that's the unique sales proposition, at least currently, of what we're getting with the products like ChatGPT and just, you know, generative AI? Is that automation piece?

Currently. The interesting thing about AI is that if you break down what is AI, because I think again, we need to start from a position of knowledge, we need to help teach people. When we talk about artificial intelligence, to me we're talking about 3 different things together. The first aspect of that is we're talking about the concept of a piece of software, because AI is just software. It's an app like any other app. The difference is, is that that app, we've taught it to think. Now, what do we mean by that? The simplest example of teaching an app to think is to understand language. If you go back even just 4, 5, 6 years ago, the idea that you could control your phone or your house with voice was absolute science fiction because software couldn't understand the spoken or written language. You used to have to, you know, type things into your Google search engine in very convoluted ways, even as little as 5, 10 years ago, because it didn't know what you were actually looking for. That first process of getting the software to think, getting the software to understand, was the beginnings of AI. Still very limited because at that point a human operator, a coder, had to teach that AI everything it knew. We then went on to the second wave, which is machine learning, and that was the idea that if I simply gave the AI a bucket of data that the AI could learn to think better using that data without any input from me, without any labor. Back to our Adam Smith question, that AI is now self-improving, doesn't require resource, it doesn't require capital, and it doesn't require labor. It just needs time and data. So data, again, back to that being another pillar of our society. And what that did is that exploded the ability of AI to understand language, imagery. So today, for example, you go into your Google Photos and say, show me photos of a cat, and it can search through your photo library and show you cat pictures because the software can understand what a cat is. It can understand what a cat looks like, and it can now go find it on its own. And then we get to the third generation, which is where you say things like ChatGPT and the generative AI world. And this is where we start to look at software that can not only understand and learn— those are our first two, the ability to understand, the ability to learn— but we can also now have the ability to create. And that's the software's ability to take those first two and go, I know what you want, and I've gone out and learned a bunch of related material and knowledge, and I'm now going to create you a picture of a cat that didn't exist before. That does, in a sense, apply to automation, and that's what people are using it for today. But I think, again, they've missed the trick because you don't need creativity to automate. Creativity is to problem solve, is to innovate, not to automate. So why are we teaching these AIs to create if all we're looking for is automation?

It's a good question. Well, okay, so you mentioned about not knowing equals fear, knowing equals opportunity, which I really love that. So are we saying that are we being, you know, too conservative with our approach to AI right now? Is there any, like, from your perspective, you know, bold moves companies should be doing with AI to just jump, you know, or make a big leap?

I think there is. And here's an interesting thought for businesses looking to employ AI. Once you realize that sort of step 1, step 2, step 3 process that have created an AI, with the ability to think. I've now given it the ability to learn, and I've now given it the ability to create. If we look at that AI as we would a little human, it's kind of how we raise children, isn't it? That first we give them the ability to think, we give them basic language skills, we then teach them how to teach themself. They then start expanding their knowledge on their own, and they start to create. The opportunity is, in many respects, if you were raising a child, you wouldn't necessarily teach your child to think and do exactly as you do, certainly not at that point in your development, because the world has changed. The world is a different place than when you were a child. So, you're trying to impart lessons to that child that you don't particularly know yourself. You know, I've got a 20-year-old son myself, so I've been through this journey. And it is always interesting when you're trying to teach your child something that you don't know, but you recognize that knowledge is valuable, that knowledge is powerful, that skill is powerful. I want you to learn it. And they go, oh, well, that's great, Dad. Can you teach me? I can't because I haven't learned it, but I want you to. The companies that are going to make the most out of AI are the companies that are going to look to that AI exactly that way, as that toddler, and go, what I want isn't for you to do what I used to do. What I want is for you to ask me if there's a better way to do it, and then go think about that. And propose back to me that, well, actually, yeah, why were you doing that? We could do this instead. It would be so much more efficient. It would be so much better. It would create so much more value. That's a very different conversation than I hear people having about AI today. You know, this whole point you mentioned, right, the AI are coming to take my jobs. I wrote an article not that long ago, and I likened it to the Industrial Revolution. I said, you know what, when they invented the steam shovel, the steam shovel wasn't coming to take your job as a digger. It just wasn't. It was just a new tool. And if you were a clever digger, you learned to operate a steam shovel, which meant you could now do the digging of 20 men. The other 19 men might not have been diggers anymore, but you know what they could have done? They could have learned to build steam shovels. Or to repair steam shovels, or to provide services to the guy driving the steam shovel. So there were whole new jobs that were being created by that technological innovation that didn't exist. And there was a great study from some of the people at MIT a number of years ago, and that's exactly what they pointed out. They said, if you look at the 1990s, early 2000s, people at that time, they were saying the same thing about computer technology in general. Computers are going to take our jobs. 10 years on, they went, did computers take anybody's jobs? And the funny thing they discovered was that 60% of the jobs that currently existed simply didn't exist prior to computers. So the computers didn't take jobs. The computers created new ones. They created far more jobs than they displaced in the same way as that steam shovel didn't take the job of digging ditches. We still dig ditches today. We're just a lot more efficient at it. But what it did do is create a whole world of new jobs around supporting and maintaining a steam shovel. Mm-hmm. So if we look at AI in that same respect, we need to stop talking about the AI taking our jobs, and we need to say, which jobs is the AI going to create? Jobs that we haven't even imagined yet. Because I want to ask that AI, is there a better way to do this? Don't do what I do. Give me a better way to get to that value, to get to that outcome. And that's going to be magical when we start to see this generative AI helping us create and becoming the better tool. That's going to change society. And the people that realize that early, they're going to be getting in on that gold rush and You know, we talked about that whole idea of when, when a new resource is discovered and staking your claim. Well, the shovel analogy is apt because who made the most money during the Gold Rush in California in the 1800s? The people that sold the miners shovels. They actually made far more money than the people digging the gold did. So here's the question that businesses should ask themselves: do you want to be the person using the AI to go dig for gold, or do you want to be the person building the AI and creating the system that people are using to dig for gold? I'd argue that's far more lucrative.

Okay, I actually can see that very easily as well. That's a very good shout. You had some real thoughts about this, Stephen. I mean, it's very clear. So give us your insights of how the next wave of AI technologies looks like in your mind.

So in my mind, if you think about any, any knowledge worker economy, medicine, law, finance, you know, anything you can think of, how do big global firms compete? Well, they compete by having that star talent, don't they? If you want to be the top architecture firm in the world, it's because you have the best architect. There's no shortcut around that. You have to find, foster, and develop the greatest talent. That's not to say that the firm that has less talented people is a less efficient business. But they will be less successful. It's kind of a cruel irony. We see the same thing in professional sports. The teams that generate the most ticket sales, not necessarily the winningest teams, they're the ones who have the superstars. So if you want to build a top-performing team, you go hire superstars. When we, when we think about what this next wave in AI is going to be, ChatGPT Well, you can kind of think of them as an agency worker, can't you? Mm-hmm. It's not a superstar. They don't work only for you. Everybody has access to the same worker if you equate them to, you know, a new provision of labor. So you're not going to gain anything by being the guy who most efficiently uses ChatGPT. Is that going to make you a world-leading company? No, because anybody can use it. The next wave is going to be the company that says, actually, what I'm going to do is I'm going to be the first law firm to design and train my own AI who works only for my firm. And I'm going to make sure that, just like raising our kids, that mine gets the very best education, the very best food, the very best environment to live in, and lots and lots and lots of security. So that nobody can interfere with it. Nobody can steal my very important data. That's going to be my data, and I'm going to use that data to further grow and develop my AI. And so you're going to come to my law firm because not only do I have the best lawyers, I have the best AI. And when I put those together, that's going to make me the next wave successful. And that other firm that goes, oh well, my lawyer is a little bit better than your lawyer, you go, yeah, but is your AI And what's the marketplace going to do? The market will realize that there are some unique benefits that AIs can give us in areas like law, like medicine, like finance. And they will go, well, I know you might have the best people, but you no longer have the best AI, which means actually I'm going to that other firm over there. So it's a whole new competitive advantage that I think companies can learn to develop internally. And if they do it right, it's going to change the market dynamics really considerably because while there's quite a lot of constraints over always having the best talent, I think a lot of those constraints won't apply to having the best AI. We're going to see countries and organizations that previously would have struggled to compete suddenly be extremely competitive because they've invested wisely in AI technology And although they might still not have the best talent, that net effect of having a fantastically developed AI with some pretty good talent is actually going to pull them ahead of the firms that have the best talent.

So what you're saying, Stephen, is that the edge of AI will be a bigger edge than a human skill.

I think it will be, you know, and if for no other reason than the earlier points I made that it's It's something you can exploit over and over and over again. You know, if you're trying to grow your accounting practice and you have the best auditors in the marketplace, they can still only do this number of audits per day. After that, they're tapped out. The AI combined with the right data can scale infinitely. You can be super global. You don't have that hard constraint that you would with traditional labor and resources.

Okay, so that puts the data in the epicenter of all of this. So ownership of data, or where you mine for data, will be, you know, the new oil era, let's say like this.

I think so.

Yeah, it's a very nice vision. How much time do you think we need to actually be in that space? Is it like 5 years, 10 years, 50 years?

I'm going to say 5 years or less.

5 years or less.

Okay. One of the things that we've seen, and this has been happening for the last 100 years now, is that the speed from a technology being invented to a technology being ubiquitous has been accelerating. There's a really great graph put out by the New York Times, I think mid-2000s it was, and they were showing how, you know, you go back not even 100 years, and those adoption curves were very sort of, you know, they were gradual. The last few adoption curves have been nearly vertical, exponential. They almost don't show on the graph. It's just a line straight up to say that, well, we invented this technology and 24 months later it was integral to everyone's life everywhere in the world.

Okay.

I can only assume history being the greatest predictor of the future that AI is going to be on that nearly vertical curve. We're talking about it in 2024. I think 5 years from now, if you don't directly employ AI in your line of business, you'll be seen as a laggard. It'll be absolutely overnight, which unfortunately is a narrow window of opportunity because there's quite a lot to do in order to build, create, curate, develop, and support these AI systems. You know, they're not going to be simple pieces of work.

Well, that, that's where I was sort of waiting to challenge you on this is like, if we're saying 5 years, we basically saying that the computing power, you know, the ability to mine the data or access to the data and as well people skills to actually make sense of that data and train AIs is already there. You don't think that we need to actually develop or have more AI specialists, data scientists, or, you know, data people in general to be able to have that fast pace? Or if you think like infrastructure speaking, people skills, you know, cloud computing, everything is there, is just about that opportunity is already present and people, if they would jump on it we basically have this in 5 years?

I think we have the prerequisites in place. Do we need more people to focus their, their skills and their thoughts in this area? Definitely. What's going to change is that obviously a lot of this AI and software-based technology, we're going to find ways for the technology to develop itself. That's going to be one of the things that really dramatically accelerates this. So if you imagine the computing power being on that sort of Moore's Law that we say, well, it's roughly doubling every 18 months, that's held surprisingly true. You know, he didn't know it at the time. It was very much a finger-in-the-air sort of guess, but we're certainly seeing that now. And because of that pace of the underlying infrastructure capability growing so rapidly, it's an exponential curve. If we double where we are today in 18 months, that's a phenomenal amount of compute and storage at our disposal if we're willing to use it. And I think that's kind of the question is, you know, if, if you're in industry, you know, you're not a tech company, are you thinking about, I need to go out and start recruiting some AI developers, I need to start recruiting some data scientists, I remember a quote from a major British insurance company, and I met them a number of years ago at a conference. And he was saying, you know, well, most people think we're an insurance company, and it's obviously, you know, it's literally in the name of your business. And he said, we haven't been an insurance company for years. We're a software development company. So 80% of our payroll goes to developers because the insurance part is so easy now. The number of underwriters we need, the number of insurance people is minuscule compared to the people that run our website and our claims management system and our quoting system and all of these other pieces of software. And I thought then, that was 10 years ago, and I thought, that's transformative. When we put AI into that mix, that transforms so many more industries, so many more places that are going to need to think about that and go, Am I recruiting the right people in 2024 for the business that I'm going to have in 5 years?

That's a good open question, isn't it? Um, so you mentioned the pace of growth of technology, and you mentioned as well the pace actually outpacing the value creation. Can you explain what you mean by that?

So if you look at why a technology exists in the first place, I think there's probably two fundamental reasons. Number one, being an engineer myself, a lot of technology exists because it can. Engineers are curious people. We look at a situation and we go, I wonder if I could do this. There's often very little thought into the should, right? Because the motivation is quite literally, is it possible? Now engineers will get very excited by things they determine are possible and go, because this is possible, I therefore attribute value to it and you should buy it. That's not quite the case though, because value— we have to get back to Adam Smith's thing, right? Are we putting less in than we're getting back out? Are we creating value? Ultimately, when we talk about value, we're talking about monetary value. And there's been challenges with the IT industry, hasn't there? Because what we've seen is an industry that in the 1970s was almost non-existent to one today that is a list of pretty much all of the trillion-dollar companies work in the IT business. And you go, that's a lot of money. If I subtract that money from the economy is the economy that much bigger today? Have I justified those trillions and trillions of dollars in IT? And for a lot of organizations, unfortunately, the answer is no, not really, right? They've spent a huge amount of money on IT, but they don't necessarily make or sell more t-shirts or cars than they did 100 years ago. And the ones they do make and sell are not necessarily more profitable than they were 100 years ago. What they've done is they've simply shifted a bunch of their expense from resources, from labor, into technology. But the net value, it's not there. And that's, you know, back to our Adam Smith conversation. That's where I think, you know, we've got to start looking at data and AI as being two of these other components to say, well, these are places you can put money in that you have to directly measure the output from and go, I've spent a whole bunch of money to store all of that data, and keeping the data, that's going to cost me money every year just paying for that storage to keep it there. Is it valuable? Show me, show me the value in that data. If it's not, get rid of it, you know, don't keep paying for it. You wouldn't keep paying for staff that weren't doing any work. You wouldn't keep paying for a building that was vacant. Well, why are you paying for IT that's not actually delivering value? And they would go, well, because they don't understand how the IT creates value. Well, great. Let's start there. Let's start to understand how that IT is going to create value to make sure that we're getting the math right. Because otherwise, what we're going to see, and, and I think it's almost inevitable, that we will get these businesses that they're going to invest a fortune in AI. They'll have listened to some of what we talked about today and gone, that's a great idea. I clearly need more AI. I'm going to go hire a bunch of software developers. I'm going to build this massive data set. And then 5 years later, their chain of coffee shops is going to go broke because they spent all their money on AI. Well, they didn't ask how is that AI going to either A, sell more coffee, or B, make that coffee more profitable. If I can't answer that question, why am I doing it?

Hmm. Wouldn't that by default create a bubble? So, because you, you, if you create, if, if you're putting money in and you're not making more, you're effectively likely seeding this with, with debt or, you know, somehow else. Because like you need the money for the IT and if, if you're not extracting more value, you are effectively creating a bubble which is driven by debt, isn't it?

Yep, and you're 100% right. Is there going to be an AI bubble? Definitely. We're going to see a lot of businesses spend a lot of money and go broke as a result because they didn't do that value math correctly when they went down the road of investing in this new wave. We saw the exact same thing in the 1990s. All of the, you know, all of the retailers in the world were going to transform their lives by having a website. What we came out of it with was Amazon, not really a lot of other people. Uh, did Amazon get it right? 100%. Was their mission to replace bookstores with a website? Not really. They looked at how digital technology could change the buying and purchasing of those assets, not just how we could do the same thing that bookstores did before but online.

Yep, that's good shout. Yep. So, um, we have this question, Stephen, which is in the spirit of the show, and it goes like this. Um, one factor, uh, in your view, which is the most critical to any tech venture growth? Only one.

Only one factor?

You can choose only one.

Yeah, security.

Security. Can you give us a little bit more?

Well, what we just talked about, the value of your data, the value of your AI, is all going to come down to your ability to protect it. So one of the things that I used to tell people all the time, people would be very, very anxious about the idea of moving their data from what they— what at the time, this is, you know, 10 years ago, would have been a server room down the hall in their office building and we were saying, you know, you really need to move that data into a proper server that's in a proper data hall at a colocation center, in a cloud provider, something of that nature. And they'd go, oh, that's, that's far too scary. My data is very important to me. That's why I keep it here in the building. And I used to say to those people, and I would say, okay, I understand where you're coming from. That, that's, you know, that's a very valid point of view. And, uh, 'You know, I like what you're saying. Before I go though, can you show me your safe with all of your money in it? Because I'd like to see that many millions of pounds in one place.' And they would look at me funny and they go, 'What are you talking about?' I said, 'Well, but you just said your data is very valuable, so you keep it in the building. So I assumed that, you know, the, the £20 million valuation you have on the stock market is in the building somewhere.' because money is also very valuable, right? And they would go, of course not, I would never keep £20 million in the building, it's, you know, it's down the road in the bank. Well, why is it safer for your money to be in an institution who specifically specializes in keeping your money safe than it is for your data to be in an institution who specifically specializes in keeping your data safe? And they would go, "Oh, I never thought about that." And I would go, "I know you clearly didn't because I'm not disagreeing with you that your data is immensely valuable. I'm also suggesting that you are in no place to protect it because you don't know enough about protecting your data." Well, unless you employ some security experts, but again, you absolutely— Are they as good as the ones that work for the bank is the question you always have to answer.

Exactly.

So I think that going into this stuff and, and what I was saying about that increase in cyber issues and where we're seeing, you know, the first potholes in our digital infrastructure, we've got to start breaking free of the DIY mentality. And organizations have got to start to realize that I need an ecosystem of partners around us. I need specialists in each of these different areas because The security of my data, of my AIs, of my systems, that's going to be the make or break. Because if my competitors can come in and steal everything, I don't have it anymore. Where's my unique value? And worse yet, I think what we're going to see, and we've seen some of this already, there was a case the other day where a courier company deployed a lovely new AI to do customer service. And within a matter of days, it was confidently telling people that they were a really terrible courier company and that people should actually ship their parcels with this other company. And they went, what? Yeah.

So did someone hack like the dataset or like how that happened?

Well, the interesting thing is that it didn't even need somebody to hack the dataset. The AI was actually able to just go read the honest reviews out in the marketplace. And formed the conclusion that any logical thinking being would, that actually we're a really terrible courier company, and then started telling the customers that. That AI needed to have some security around it. It needed to be a little bit sheltered so that it could at least spin the truth and deliver good customer service without maybe necessarily being that level of honest, because that's not what we expect of customer service. The customer service is supposed to make the best of the situation we've got, not tell the customers, sorry, we're really bad, why don't you go deal with that guy. These are all security concerns, and organizations today, they're just not set up to deal with security for digital infrastructure. Um, even in the service provider space, some service providers are good at it, some are not. The shameless plug for our business, why we exist, We're architects. Just like in the construction world, you wouldn't hire the brickie down the road to design your office building. You also wouldn't expect your architect to roll up his sleeves and start laying bricks because they're different skills. We need both of them. In IT, we've got too much of the DIY mentality that sort of says everybody's a jack of all trades, everybody can do this stuff. It was true, it's not anymore. The body of knowledge is too vast, the complexities are too great. We now need to start realizing you have to have all those specialties in place. You can't expect your brickie to do your plastering. He can do it, but your walls are going to look awful because he's just not a very good plasterer. He's an amazing bricklayer though. So build a team, build an ecosystem, and work with partners. Don't try to hire these people in internally because unless you are a hyperscaler, a Google, a very specialist firm, you're not going to be able to hire the best talent. The people that you're getting are the people who couldn't get work at the more interesting place.

Yeah, so it's a buy versus build decision.

Exactly, yeah. And that's not to say that you shouldn't be designing, that you shouldn't be building, but build the part that creates value. Don't build the uninteresting stuff because the uninteresting stuff is just that. It's uninteresting. There's no value in it.

Exactly. All right. So last question, Stephen, if you would give advice to businesses right now, you know, looking to maybe making a step towards AI, maybe they already playing with AI, with your visions around the future, how that advice would sound?

So I would say that what businesses should be doing is looking to how AI can, in the short term, eliminate waste in their business. That's going to be the, that's going to be the quick win. So things that are particularly time-consuming, get an AI to do them. What you want to do then is not get rid of those people because that's the, oh, AI is coming for our jobs. What you're trying to do is to free those people's time so that you can get them to think about what the AI is going to do next and start to plan what the transformation of your business is. There's people that have coined the term Uberfication. When Uber started out, Uber was not a taxi company, was it? It was a software company. They didn't own any taxis. But what did they do? They transformed the taxi industry. The taxi companies at the time were focused on resisting Uber rather than asking themselves the question of, well, is this the opportunity for an existential change in how I run my business? Maybe I don't have to buy cars either. They're not buying cars. Maybe I can find a new relationship with my drivers, with my local operating areas. And that requires people who understand your business. This is what people often forget, is that they'll go out to a service provider or a tech company and they'll ask the tech company the hard questions. How can your tech help my t-shirt business make more t-shirts? Microsoft doesn't know. They don't make a lot of t-shirts. Don't ask them the hard questions, ask your people the hard questions. But use the AI technology to free your people's time so that they have that mental bandwidth to think about your business and to think about how your business might transform. And then cycle that back into the AI to then see, well, can the technology enable that transformation? Can it support my transformation? And now is the best time to invest in those kind of decisions. Because we're still waiting for that rate, waiting for the wave to crest. You know, if you wait until the wave is crashing down on top of you, there's going to be so much competitive pressure, there's going to be so much more pressure for the talent that you can now hire and train and develop today, that it's going to be much, much harder in 3 or 4 years to survive that transformation because you'll have to do it far faster, it'll be more expensive, and you'll have a lot more competitive pressure against you to do it. If you start that transformation today and going, there, there's a way that I can free time in my business so that I can get my people to focus on my business. And then I'm going to create the ideas for how to transform my business. And I think that's the winning formula.

I love that, Stephen. Thank you very much. Anybody listening, if you want to talk about AI, how you can apply AI security, and a lot more, we Stephen Harrison, who's a co-founder of InfraTech Digital. Reach out, the details are in the show notes. Thank you, Stephen, for being with us today. Thank you for all the knowledge. It was great. Thanks for attending.

Thanks very much, Mike.