Digital twins: The next frontier of public service delivery

Bill Eggers: This is Government’s Future Frontiers, the podcast from Deloitte that asks questions today to help create tomorrow. I’m Bill Eggers, Deloitte’s executive director for the Center for Government Insights. This episode comes to you live on location from Barcelona at the Smart City Expo World Congress, which brings together key players from all over the world.

So, it’s an ideal place to explore this episode’s subject—predictive digital twins and engineering tomorrow’s infrastructure. A digital twin is a live virtual representation of a real physical asset in the world. It could be a building, a power grid, or even an entire city. Creating a digital twin provides the ability to try things out in the virtual space before actioning them in the real world.

So, let’s get stuck into this subject with two fantastic guests, Justin Anderson from Connected Places and Nick Holmes from ServiceNow.

Nick Holmes: Well, first of all, thank you very much for having me. Awesome to be here back in Barcelona, as always.

My name is Nick Holmes. I am a veteran of [the] government [sector] and [have been] working with government agencies for about 23 [or] 24 years, or so. I work at ServiceNow. I’m based in Dubai. I’m a global public sector director of sustainable infrastructure and transportation. So, I spend a lot of time at conferences like this, really trying to understand what the leading practices are; what are the things that our platform, from a ServiceNow standpoint, can do to help our customers to solve their problems.

Eggers: Terrific. Justin?

Justin Anderson: Well, again, thank you very much for having me. I’m Justin Anderson, managing director at Connected Places Catapult. The Catapults are part of a network funded by [the] UK government, and we’re here running the UK Pavilion. It’s great to be back. I think this is my 12th year here. I love this show. It’s a great place to meet and see people that you haven’t seen for a year often.

This is a great environment to see the trends and the changes that have happened. Sometimes, not that much happens over the course of a year, other than bigger stands for some countries and a few advances in technology. Connected Catapults are focused on bringing together cities so they can learn from each other. And we have a very strong focus on ensuring that we can align the policy intentions of UK government with the country and with the markets. So, we sit in this interesting space that makes sense of the policy, and it helps guide industry so that it can make the most of it.

Eggers: Wonderful. Well, let’s get into the subject, and Nick, it’s great to have you here. We collaborated on a big 250-city survey, AI-powered cities, which I think is the largest of its kind in the world, and hope to be doing round two very soon. Now, the subject here is digital twins. Both of us have been walking around, and it seems like you can’t go more than about 5 yards without seeing a different digital twin or hearing about it in a conversation. Everything from wildfire prevention to transportation planning to workforce planning—just so many different digital twins, and it really seems to be one of the big themes and trends of this year’s conference. Tell us a little bit more about digital twins and why we’re seeing such an explosion at Smart City Barcelona.

Holmes: Yeah, I think that’s a very true observation. Like you said, as we walk around, we just see so many different examples, and I think it really comes down [to this question] in my mind: Why are we seeing such a prevalence of them right now? I do think it comes down to our good old friend, the data.

To drive a digital twin, which is really just sort of like a replica of—it could be your environment, it could be your building, it could be your city, you name it, like you mentioned—a whole bunch of different use cases, you need really good data to be able to do that. And the idea is that you can run what-if scenarios. So, you can play around, and you can start thinking about, well, what if this happens or what if that happens? And I think that’s very powerful when it comes down to decision-making because we’re not just making decisions in the dark.

I think that the inflection point [here] and why we’re seeing right now [that] digital twins are becoming more popular is: One is that accessibility of data, being able to capture the data. When I first started out doing smart cities, we were [at a] very early stage. We didn’t really have that many IoT (Internet of Things) sensors.

You know, really the big pivot point there was the cost and also the battery life of all things. If you had a sensor and you had deployed that sensor, and it ran out of juice, it stopped communicating. It stopped giving you the data. And now I think what’s happened is we’ve figured out those issues and those problems.

Now we’ve got great sensors providing a lot of data, a lot of information. And now I think we’re into the next phase, which is: What do I do with that information? How do I process that information? Because if you’re not making a decision from the digital twin, then why are you really doing that?

And then, as we were talking a little bit before, I feel like we’re at that same sort of stage where a couple years ago we just talked about AI, AI, [and] AI. Everything was AI-oriented. And we sort of got that, and the study actually came out that you, you were alluding to, and said, one of the findings was people were doing AI just for AI’s sake.

At ServiceNow, we have a digital workflow platform that has AI baked into everything that you’re doing. So, you’re not really thinking about consciously, “Hey, I’m doing an AI addition onto this.” But similarly, could there be a process fix? Could there be an organization change? Could there be some sort of other way that you can actually sort this out, versus jumping down the AI route? And I wonder if that’s also the case here with digital twins. I think there are very good use cases—very sound, solid use cases. But does everyone need to have a digital twin in their back pocket? I think not.

Eggers: Right! Are we at the top of that Gartner hype cycle or not?

Anderson: Can I just pick up on that? We’re past the top of the hype cycle. I think we’re going down the slope. But I think when we get back up the slope, it’ll be much bigger than we could possibly imagine.

One of the questions that I often get asked is, “What is a digital twin?” And the first thing I’ll say is that “it’s not a noun, it’s a verb.”

This is a journey that many organizations have been on over a period of time. You can’t go out and just […] Well, if you do, you’re making a mistake.

You can’t just buy a digital twin. There is no one-size-fits-all. There are many, many different flavors, and herein lies the real challenge, which is, that you have to make them to fit the purpose and the use case that you’re trying to build those scenarios and decisions, and the what-if statements that you’re looking to try and answer.

Eggers: Now, you mentioned, Justin, that you think that the future is going to be a lot of use cases and things that we maybe can’t even imagine that are going to be very transformational. Could you give us a quick peek into that?

Anderson: [That is] What they are? [You mean] the ones we can’t imagine? I’ll try. Well, first of all, one of the challenges is that we’ve been building digital twins in silos using different standards and frameworks. And a twin that Nick might be building and a twin that I’m building may or may not easily communicate. And it’s not just the technology that is the challenge; it’s the governance that sits around this. So, we don’t know yet what will happen when we start to bring these twins together as autonomous decision-making agents.

Once we start to connect different data sets that essentially represent different sectors or different ideas, and merge them together, we will create something that is difficult to comprehend. What I would suggest though is that as we build that digital nervous system around the planet, what will emerge on top of that will be something that would be like looking back a hundred years and seeing horse and carts on the street, as to where we are today.

Eggers: Right. Well, Nick, I’d love to get your thoughts on that, given that ServiceNow is doing exactly what Justin is talking about.

Holmes: Yeah, I’m very much in favor of this sort of concept of creating a library—a library of digital twins. I mean, we’ve got to get the governance right. We’ve got to be interoperable. But wouldn’t it be really good? I mean, I’m based in Dubai, but I spend a lot of time in Africa. And we’re talking about how Africa is growing, and Africa’s growing so fast. One of the terms that I hear a lot—I’m sure you do too—in Africa is the concept of leapfrogging.

Wouldn’t it be cool when Africa is able to have all the infrastructure pieces, they can leapfrog so much ahead because they don’t have all the legacy things to do? Well, what if we could have a marketplace with digital twins and playbooks that we could pull down off the shelf?

It’s not going to be an exact fit, but it would be a close enough fit to get started. Then you’re not starting with that blank sheet of paper. You’re moving on from that. I get excited when we start thinking about the interoperability [of the] pieces like that.

Anderson: I like that. I 100% agree. The UK government has committed a hundred million pounds to building what we call a national data library. And that will be all about discoverability and whether it’s actually discovering a full twin or the data sets that are available to make it easier for that connection, it’s recognized as an important enabler for our industrial strategy.

Eggers: What kind of data are we talking about that are needed to build the digital twins? IoT data, obviously. What are some of the other kinds of data, and how do you build those into a simulation? The other thing I’ve heard a lot about is that if the data is bad, you’re going to have a lot of very bad predictions, in terms of the twins. So, let’s talk a little bit about that data piece.

Anderson: So, there’s a plethora of data that feeds digital twins that might be static reference data that is historical in one format. So, maybe you’re pulling it out of a system that’s been running for a while—a legacy system—and you have to extract that. Supplement it with IoT data. That could be sensors; it could be video that is pulling information straight out of the streets, and we’ve got to merge that.

I think there’s a lot of work right now that’s being done in AI around VLMs (visual language models), which I think is one of my big takeaways from the show is how we understand the analytics behind them.

Eggers: And VLMs, could you explain those?

Anderson: Visual language models. So think about ChatGPT. But rather than it being about text, it’s all the visuals and how you interface and interact with real images in real time and create scenarios that you can play out very, very rapidly, and simulating the real world off the back of it. Let’s see what happens and see how people move around.

Let’s just take an example: Let’s say we have some sort of an event that requires an evacuation. We can simulate the evacuation—recognizing the people in a city [and the] movement of the people across transport networks through roads and across trains.

Working out how perhaps the power is going to be impacted as a result of the evacuation: Start bringing that data together, simulate it using the visual language models to watch what’s happening, and feed it back and analyze it, and in real time, make decisions about how we should deploy our emergency services.

Holmes: Yeah, I think that’s one of my big takeaways, as well. I mean, back in the day when we were trying to create those sorts of visual models, the amount of training data, the amount of custom tagging you had to do, you know, I was looking at some of the demos we’ve seen here. I mean, [with] those VLMs, it’s just plug and play—off you go to the races.

I mean, that is such a game-changer in my mind. And I do think, going back to the question that you asked, we obviously want to have the best data we possibly can. I think about it in a couple of different ways: One is like if you have a platform like ServiceNow, we can very much tie into all of those sort of legacy systems. We’re not necessarily looking to replace your entire legacy infrastructure, legacy application real estate. That’s just too complicated. People spend too much time, too much effort. No one wants to be told that their baby’s ugly, right?

And that you can’t use the data. So, being able to use existing data sources, plugging in those new models, which I think can smooth over, to a large extent, a lot of what’s going on, I think that becomes absolutely critical to what we’re looking to do.

Eggers: So, what about some real-use examples we might go through? One of the things that I’m really interested in is we’ve got the Winter Olympics coming up. We’ve got [the FIFA] World Cup coming up, and then we’ve got Summer Olympics in Los Angeles where I used to live; and sports has been a leader over the years in digital twins and using vision, computer vision, and so on. But with the Olympics, you both have the athletes, the user experience, you’ve got emergency management areas, you’ve got security infrastructure, so many very complex things that all have to come together. Would that be a great use of digital twins?

Holmes: It’s so funny you say that, and this is a shameless plug for your booth, and what you are basically showing over there. A colleague of mine, Mick, and I saw that and we are thinking along very much the same lines. We’ve got the events, the events management, of course, the in-stadium experience. ServiceNow works very closely with [the] NBA, with the Hockey League, and all these big sports things.

I know you guys do too at Deloitte with the IOC, and what we are thinking is, to your point, it’s the visitor experience. It’s the fan experience, but also the employee experience. I think the statistic I heard for an Olympics games is the temporary staff [number] is around 40,000 that need to be attracted, retained, hired, trained, put into the right position, told what to do, told how to do it—and that’s a massive, complicated headache for someone.

What if I could simulate that? What if I could simulate the causes and effects? And that’s just one area. So, on your board where you can move all the pieces around, just thinking of all those different areas that you would have transportation, customs and borders, police, the tourist experience—there’s just so many verticals. That would be one really, really good digital twin.

Anderson: Can I just add to that thinking, which is the opportunity to deploy state-of-the-art digital twins and the technology that will sit underneath that—but also an opportunity to ensure that public and private sectors work well together. Because if you are bringing a large number of visitors into a city, then the city itself needs to react to that. The stadium is there to be able to bring together those visitors, and managing that visitor experience is absolutely key. But you want to make sure that the safety around the city as well.

And that interplay of data between the city and the stadium, which may well have a number of different private entities, managing different parts of it is key. And too often, you’ll find that the city has got data sets about the transport systems, about its services that don’t connect into the private sector. So, I think where you’ve got that opportunity to and critical [the] mass to create something that really needs to be delivered in an exceptional way, it also creates that tension that will drive that public-private sector interoperability that is key for us.

Holmes: I have another really good use case that I had heard a while ago. And talk about digital twins, I think, it is an interesting one where there’s an interplay between private sector and public sector—and that’s in the world of government procurement.

We know government procurements are long. They’re lengthy, especially when we talk about infrastructure projects and those kinds of things. What if you had a digital twin that then enabled the stage gates of payments and completion timelines and completion deadlines? We all get frustrated when we drive down the road, and the road is closed because there’s work going on, but you don’t actually see any workers. And of course, as citizens, we’re sitting there saying, “Well, what’s going on?” But there’s an interplay there: What if we could do a digital twin of the procurement and then make sure that the vendor is delivering on that? I think that would be a good use case.

Anderson: I like that. One of the challenges that we face is that there is this drive to want to share data or provide access to others to use that data so that we can create something better than the sum of the parts. But the problem often is: I don’t want to share my data. You don’t want to share your data. The last time anybody around here shared their data, there was too much risk and we had a problem. So, we’re operating in an environment where, in fact, we’re disincentivized to do that. So, some kind of mechanism that allows for better procurement and incentives to be aligned. Some kind of data exchange that ensures that actually when I do, there’s an upside to this.

That transaction is recorded. It can be assured and audited. And we can build the trust in the system that then builds the critical mass to the level where, of course, I’m going to be sharing, I’m quite happy to do it. Now we’ve built that trust that lets the data flow. And if we create additional value and I was the one that had the source data, I may be rewarded at a later stage. Far too often, that doesn’t happen.

Eggers: I couldn’t agree more. My most recent book [Bridgebuilders] is on public-private collaboration: We have a whole chapter on data as the new language, and it’s around sharing the data between public and private sectors [or even] universities. Because we talk about data silos within government, but the private sector has so much of the data now, and we have to figure this out. Now, with all the talk about digital twins, a question for both of you: In what cases does it not make sense to have a digital twin, to go down that route? Are there any or not?

Anderson: So, let me just jump in with one digital twin in the United Kingdom, which is a fantastic digital twin: It’s Kraken, owned by Octopus Energy, that now is responsible for balancing 50 gigawatts of energy. It has 60 million customers in 26 countries around the world. And the twin that’s being created by the company is arguably worth more than the company.

Eggers: How so?

Anderson: Because of the value of data—exactly the point you were making. It’s because they are able to drive those connections; but importantly, [they] help balance the grid.

And so while it’s bits and bytes that are moving around, they are related to energy. And it’s that relationship to the electric current that is key. The value sits there. It’s modeled it; it works out how to get it flowing, and therefore, the economic model that underpins it—it drives the adoption and drives the use.

Holmes: I think I’ll keep it really simple. I think it boils down to the return on investment that you’re going to be getting by building your digital twin. Like you said, sometimes, you don’t need such a big, complicated initiative to solve a simple problem. If you’re getting the output that you really want to be getting from it and it’s not costing you that much to be able to figure out how to build that digital twin, maybe you can resurface some of the information that’s already available.

Maybe pull down one of those playbooks that are already existing, then sure, go for it. But I do think that to the point we were making before and the hype cycle, just before you jump in the water, just make sure that you really know how deep that water’s going to be and how long you’re going to [take] to get there.

And that it doesn’t distract you from the actual goal that you’re trying to accomplish. Because if you’re doing that, then you’re just spending cycles. You’re going to get frustrated, you are going to lose your stakeholders. The change management aspect goes out the window. You’re just going to not look so great.

Eggers: So, if we look out five years from now, and we’ll end with this: What is the wildest example you can think of a digital twin? We each have our digital twins of ourselves. Maybe not with all the brain weights, but is that something realistic?

Anderson: You’ve got a smartwatch on there—that’s already the beginning. It’s monitoring your vitals, and it’s building up a picture. Over time, it’ll look at the trends; it’ll predict whether or not, in fact, it’s time for you to get up or do something that you need to do. But if I were to look ahead, I think the key thing is that today we still think of digital twins a little bit as a mirror of the real world.

We’re able to look in the mirror; we’re able to see what’s going on; we’re able to use it to predict the future to a certain degree. I think we’ll see that shift from a twin being a representation of the world to a decision-making, autonomous piece of a broader digital nervous system that wraps itself around the planet.

And those twins are responsible for different parts of the planet and the connection, and most importantly, to pull in the data, make sure that it’s well-organized, make sure that the hygiene of the data is properly considered before it is then passed into some part of the cortex.

Eggers: I’ve read a lot of science fiction books around that, where things might go wrong in that kind of a case.

Holmes: Yeah, once again, I’m really a big believer in keeping the human in the loop. So, when we talk about these autonomous systems, I mean 100%, right? If I’m doing something relatively simple, straightforward, you know, it’s completely autonomous.

Tier one, IT help desk, for example. Password resets. Sure, go for it. But when you start talking and when you start talking like that, getting into the cortex, that’s when I get a little bit nervous. I’m like, “Are you sure that’s exactly the direction that we want to be going?” So, I do think those safeguards become very important as we start doing that.

Not to slow us down necessarily, but just to make sure that we are not having bad actors who, of course, are out there all the time trying to disrupt what it is that we’re trying to do. Several years ago, we were talking about the omniverse, and that was to me another fad.

But I do wonder, and as we’ve said, the technology is there, the data is getting to where it needs to be. It becomes a bit more of a human. Does the human want to entertain this idea and the human want to do this? I wonder if we will be walking around in this virtual universe and actually seeing what it’s like. Perhaps, we’re driving our cars around just to see how well it worked when we had an interchange where there were all these road accidents.

We’ve been able to fix that. We’ve been able to model that out. All those good use cases that have very positive, human impacts. Could we mirror those in a bit more of a graphical user interface?

Anderson: Can I pick up on the “going wrong” piece? Because this is a really important piece.

Could this all go wrong? Once we start to have decisions that are being made—perhaps, a little bit outside of our day-to-day purview. I think that is where assuring the system is key at every level. We’ve got to have the right standards that underpin it. We’ve got to have the continuous learning, continuous monitoring to make sure that models aren’t drifting in the wrong direction.

And then at the top level, we’ve got to make sure that we can assure the whole system, but it’s not going to be a single system that rules it all. It must be federated. It must have federated governance that ensures that we can still have the person in the loop, ensuring that we can control whatever it is that we’re going to be building. But governance is key, and we’re going to need governance to move at the speed of data.

Eggers: My thanks to Nick and to Justin for their fascinating insights about digital twins. If you’d like more from Government’s Future Frontiers, you’ll find all our previous episodes wherever you get your podcast. And to make sure you don’t miss new episodes, be sure to follow the show on your podcast platform of choice.

This podcast is produced by Deloitte. The views and opinions expressed by podcast speakers and guests are solely their own and do not reflect the opinions of Deloitte. This podcast provides general information only and is not intended to constitute advice or services of any kind. For additional information about Deloitte, go to Deloitte.com/about.