To explore how generative AI (gen AI) could transform the future of work globally, and in Singapore particularly, McKinsey partner Sanjna Parasrampuria, an expert in data, artificial intelligence and generative AI, talked to Terence Chia, assistant chief executive of the Infocomm Media Development Authority (IMDA) of Singapore, and to Kathryn Kuhn, an alumnus of McKinsey’s Digital Practice. The conversation led to the conclusion that collaboration is vital to ensure that leadership and talent are future-ready, and that companies look to capture the significant value of generative AI rather than be distracted by its power.
An edited version of their conversation follows.
Sanjna Parasrampuria: Hi everyone. My name is Sanjna and I’m a partner at McKinsey in the Singapore office. I have the pleasure today to be talking about a topic that I am very passionate about—generative AI and the future of work. While I say I’m passionate about this, I know that the topic has caught everyone’s imagination. So, to that end, I’m very pleased to have Terence Chia and Kathryn Kuhn to talk more about it. Before we dive into our discussion, Terence, could you introduce yourself quickly, and then you, Kathryn?
Terence Chia: Hi everyone. My name is Terence Chia. I’m assistant chief executive at the Infocomm Media Development Authority (IMDA) of Singapore, where I oversee the corporate group and human capital cluster. We look at how we can develop talent for digital industries.
Kathryn Kuhn: I’m Kathryn Kuhn. I was with McKinsey until recently, where I spent a lot of time talking on the topic of how we think about our workforce—especially our technical workforce—evolving as our technologies continue to evolve.
Sanjna Parasrampuria: I have no hesitation in saying that 2023 was the year of generative AI. If you look at some of the facts: OpenAI’s valuation went up to $80 billion after being about $29 billion in January 2023; 13 generative AI start-ups became unicorns in 2023 alone; and it took ChatGPT around two months to reach 100 million users—one of the most rapid user growths that we’ve seen. And “hallucinate” was Cambridge Dictionary’s 2023 Word of the Year. So certainly, there was a lot of impact in 2023. I thought it might be useful to start unpacking some of this. Kathryn, from where you sit, what are some of the key generative AI trends that you’re seeing globally?
Kathryn Kuhn: What’s interesting to me is that generative AI is not new. We’ve been building toward this for decades, whether it has been advances in advanced analytics, machine learning, or natural language processing, we are continuing to evolve these technologies in the artificial intelligence realm. What is different this time around is how we, as a collective, are responding. And, when I say “we,” it’s we as a workforce, an economy, the private sector, business leaders, as well as governments and agencies around the world. They’re all responding differently to this next leap in technology. The other thing that’s new is that there is no industry that isn’t talking about this, as well as segments of industries—whether it’s banking and financial services, insurance, healthcare, pharma, retail, consumer products, and manufacturing. All these different industries around the world are talking about the technology and exploring the possibilities that it can bring.
Sanjna Parasrampuria: Just to dig a little bit deeper into what you just spoke about: there are certain patterns around some of the key functions across all these industries that are being discussed and this is where some of the biggest transformations are going to come about. In sales and marketing, from a creative perspective, we’ll probably see a big change in how things were done previously versus how things will be done from a generative-AI perspective. There’s an element of software engineering with copilots coming in, which helps you write code faster, review it better, and be in a position to comment on code as well. So, Kathryn, are there patterns around some of the functions that we are seeing, where we feel like these are the key families of functions in which we’ll see the biggest uplift?
Kathryn Kuhn: We’re now seeing some early adopters and early patterns. One of the early adopters is the legal industry because of generative-AI’s ability to summarize documents very quickly and generate first drafts. In these types of industries, you can very quickly accelerate activities that often can take hours and do them in 20, 30 minutes. That’s the first draft. And there are other areas generating copy. For example, people who are in marketing- or marketing analysis-type functions, who generate a first passive copy for different versions of media or tailor to different audiences, can now hypertailor copy and generate those first drafts very quickly.
In the customer service realm, banks and financial analysts, for example, are experimenting with the idea that a chatbot can help generate a hypertailored script targeted at the person you’re talking to—assimilating and synthesizing the latest information, either about industries or investment strategies or what’s happening in marketplaces, and doing it in a way that is easier for someone—maybe a junior financial analyst—to get up to speed. By using these technologies, a faster and more fluent script for talking to customers is created.
Then there is computer science and software engineering. Things are turning up there. For example, if you need to move from an older programming language to a newer one, you can now generate your first drafts of code in your target programming language by taking an old block of code. Another interesting trend we have been noticing is that somebody who is not very proficient can now quickly generate first passes of data code by using generative AI. For instance, this could help in a public service agency that is having trouble retaining technical talent.
Sanjna Parasrampuria: I think one of the biggest unlocks comes with the ability of generative AI to analyze the multimodality of unstructured data across text, speech, images, and code, without necessarily requiring to label or train that data. The second unlock is the ability to sound like a human. You brought up a great example of the chatbot. Chatbots have existed for the past decade, but what’s now different is the ability to have open-ended questions, meaning that you’re not limited to a predefined response set. I was with an insurer earlier this week who asked, “So, what will be so different, Sanjna?” With them, if somebody is chatting on a chatbot to an insurance claim customer, the bot can actually review the images and the videos of the accident, look at the text assessments made by the appraiser, and provide a real-time response.
Terence Chia: I agree with a lot of what Kathryn has shared. There are two key traits of generative AI that lend themselves to the large, outsized impact it is having globally, as well as in Singapore. The first trait is that it’s a general-purpose technology and the second is that it is democratizing access to the technology underlying it. So how that manifests in Singapore is analogous with how digital technology has been spreading across Singapore and the world. To give you an idea, last year IMDA released the Singapore Digital Economy report. And, in that report, we found that 17 percent of Singapore’s GDP—that’s around $78 billion (SGD 106 billion)—is attributable to digital technology. The key is that the bulk of this is not in the information and communications (infocomm) sector nor the tech sector, but actually comes from the rest of the economy through the use of digital technology in those other sectors. It’s an important point.
I see the same phenomenon happening with generative AI and AI because of their general-purpose nature. It’s something that can help not only very technical people who know how to use the technology, but it can also unlock so much potential for others. I’m going to borrow something that Michael Chui, a partner at the McKinsey Global Institute, said recently about how AI and generative AI can give people “superpowers.” You could say there’s a bit of hyperbole, but I think there’s a lot of truth in that concept.
Sanjna Parasrampuria: I absolutely agree and I’ve had many debates with Michael on this. I guess the area where we agreed on was that all forms of traditional AI and generative AI will continue together. And, in parallel, generative AI is probably not going to look to replace any of them.
Let’s divert our discussion to another area, probably also top of mind for most organizations, and certainly governments, which is about the impact on the workforce at large when we think about generative AI, as there are concerns that this is going to cause augmentation or disruption. One can argue both ways. There could be areas where we’ll see a lot more disruption and augmentation, but largely, one could say that there is a lot more value to be derived across functions in general. Kathryn, what are some of the jobs and roles where you see some of the biggest applications of generative AI? How do you see those roles evolving and whether they are going to move to disruption or augmentation?
Kathryn Kuhn: Right now what we’re seeing is that the effect is leaning much more toward augmentation as opposed to disruption. If there has been disruption, it has been in an industry or a function that was already being disrupted and generative AI is accelerating that disruption. The other thing is that progress is not linear. You’ve mentioned “hallucination” as being the word of the year—copilots can generate a first pass of code and it can be wrong. And, if you think about the software development process, that mistake is not always immediately apparent when you are just inspecting the code. So, when are you going to catch this incorrect code as it pushes through? It’s definitely an augmentation and it’s definitely an experiment. Those experiments can be wrong and set you back as much as they can help accelerate your productivity.
Sanjna Parasrampuria: Terence, there’s a lot of conversation and proactiveness from a Singapore perspective about doing both things right, being proactive about how to prepare the workforce to leverage the opportunity and, at the same time, maximize on the value potential that the opportunity brings to the country’s and industries’ competitiveness at large. What are your thoughts from a workforce perspective?
Terence Chia: This is something that we’ve been preparing for years. I’m not saying that we had an AI or generative AI model that told us all this was going to happen but, if I take a step back and think about the impact AI and generative AI has on the country, the economy, and the workforce, there are a few different components of how we can help make the best use of the opportunities.
The first is infrastructure. AI and generative AI are built off the internet, the flow of lots of data at an accelerating rate. As early as 2006 in Singapore, we launched the [Next Generation] National Broadband Network. Back then, if I had a 56 kilobits per second modem, it was considered good. The Singapore government was telling everyone, “Hey, we’re going to give one gigabit per second to every home.” A lot of people wondered what they would do with all that additional data, but we went ahead.
And by building ahead of demand, that core infrastructure allowed us to move much faster when it came to the internet economy. Just a few weeks ago, we were talking about raising that capacity further to ten gigabits per second. So that’s one of the first precursors—having a very strong infrastructure.
But the other precursor, and I think the bigger one, is making sure that we are coordinated and have a plan. In 2019, Singapore was one of the first few countries to launch a national AI strategy, covering various projects in different areas like education, healthcare, and security, but also enablers to strengthen the ecosystem. Building off that strong foundation, we launched the second version of our AI strategy, National AI Strategy 2.0, and we will spend more than $735,188,700 (SGD 1 billion) over the next five years to catalyze AI activities in Singapore.
A good amount of that is going toward talent development. I’ll unpack this even further. First of all, when we think about talent development—the workers and the workforce—the key prerequisite is a strong foundation in the education system. This has been happening in Singapore, not specifically for AI, but because the country wants to ensure that our workforce is globally competitive no matter what trends, technologies, emergent issues, or disruptions come our way. So, we have a system with a solid STEM foundation.
The second component is industry, because you can train a very strong supply [of students], but then you need to make sure that there are good jobs and opportunities for them. To do this, the government aims to make Singapore attractive to companies, whether they’re multinational corporations coming in and catalyzing innovation or start-ups, by offering them an environment that supports business setup and expansion.
Kathryn Kuhn: How can we have a workforce that has a continuous innovation mindset?
Sanjna Parasrampuria: I’ve been living in Singapore for 16 years and have experienced much of the infrastructure changes. I ran a start-up in the AI world at a time when I had to explain what machine learning was and how to recruit talent. So, I’ve seen that evolve personally. Having said that, there is certainly an acceleration of skill set shift that is happening on the back of generative AI. I’m wondering, is there anything specific, not just to Singapore? What do people need to think about when they talk about creating this nurturing ecosystem?
Terence, you touched upon creating the right skill sets from a supply standpoint. Then there’s the other element—there are going to be certain levels of roles that will see work being done differently. And how do we equip people with the right skills that will be more future-proof? There are upskilling and reskilling pathways that we talk about in general. How are we thinking about readying parts of the workforce from a generative-AI perspective, not to just contribute to the newer world but also to understand how to consume it to better their work in the office? Terence, starting with you.
Terence Chia: I’ll break the response into three parts—first, in terms of the general ecosystem or setup we need to have; second, people currently in the workforce; and then third, people who are not yet in the workforce.
In Singapore, we’re very fortunate to have a mutually supporting ecosystem of different partners. There’s my agency, IMDA, where we focus on infocomm, media, and digital, but at the same time, there are other government agencies focused specifically on skills, workforce training, and unions.
With that as the backdrop, we can have good collaborations. For instance, we have the TechSkills Accelerator that has been around since 2016. It places and trains locals in good tech jobs—we have placed more than 17,000 locals in these jobs. And on top of that, we’ve also reskilled more than 231,000 professionals. Where are these workers being placed? They’re being placed in AI-related jobs, such as AI machine learning engineers and data scientists, but also in other kinds of tech jobs across the economy (such as manufacturing, finance, logistics, and infocomms). So, that’s one key component, which is how we deal with the workers currently in the workforce. How do we give them the skills? How do we make sure that translates into good jobs?
At the same time, we need to proactively identify what the nature of the disruption is or the impact that is coming to workers across the economy. For that, we have the Jobs Transformation Map. We have multiple Jobs Transformation Maps across the whole economy. Specifically for information and communications, we have one that focuses on things like AI, cloud, software engineering, and how we can make sure that workers continue to have skills in these areas to remain relevant.
Then the last component of my response is the students, the new workers, and the people who have yet to come into the workforce. For those, we have a strong partnership with the institutes of higher learning. Over the past five years, for instance, we have doubled the intake of technology and information and communications technology (ICT) students in our local universities, because we’ve seen the shortage of and strong demand for tech talent. Today, about one in five of these students are doing ICT majors. And beyond the quantity, we regularly review the curriculum to make sure that the quality is there and that it remains industry relevant.
Kathryn Kuhn: But how do you reshape the workforce that you have? Right now, what companies are doing first, at least in the private sector, is trying to make sure that they capture the value of these technologies. We worked recently with a client who had deployed a copilot and, after about two, three months, less than 11 percent were regularly using it. For the company, it was a fairly sizable investment to continue to license the copilot. What they needed was training for their people. Focusing first on software engineers: how do you help them create a collection of use cases, not just what can they use today, but to also create a community where they can continue to share those best practices and ideas? So, you create an innovation and experimentation culture—then you can really see the value of the technologies that are available today.
Governing it is both offensive and defensive. Offensive as in where’s the market going, what does it mean for our own products and our competitors, and how should we, as a company, respond to those competitive forces? Do we need to change a pricing strategy, for example, or do we integrate it as part of our pricing strategy? As defensive: how do we protect ourselves to make sure that we are using it well, that we can look at the data provenance, and that we understand where the data is coming from that generate the solution. Keep a watchful eye on emerging regulations and policies from the governments in which they operate. Then also, how do you respond to incidents and what does an incident response look like?
Sanjna Parasrampuria: We have spoken a lot around some of the opportunities that exist and what we can do from an ecosystem setup perspective to enable them. How do we think about the workforce and empower them to leverage this without getting disrupted in a way that they don’t know what to do with the freed-up time? Terence, are there any specific challenges that you generally see and that may or may not be in the Singapore construct? How are you thinking about ways that companies are looking to solve this?
Terence Chia: I want to build on what Kathryn mentioned earlier about culture building and data-based decision making, and how to deal with the executive group. It’s related to this question because, fundamental to all of this, is making sure that we have the right kind of people leading our companies and the workforce. One of the key prerequisites is that they have the right kind of exposure.
So, linked to my point earlier about students, beyond the fact that we’re working on quality and quantity and making sure that everything stacks up from a macro workforce point of view, we also want to make sure that we are grooming the next generation of tech leaders—so that these people have the right instincts, the right culture, the right values, to be able to bring us through the uncertain times and challenges. For IMDA, we’ve been offering scholarships to do this and, since the early 2000s, we have given out more than 800 of these for students to study infocomm media all around the world. It’s an industry scholarship, which means that after you graduate, you work in the industry. We’ve had examples of people who have gotten scholarships, come back to Singapore, founded an AI start-up, grown it, and made the ecosystem much more vibrant. We’re investing significantly so that people can get global exposure.
We agree that, as technologies like AI advance, we must find ways to guard against misuse. This can be things like cyberattacks, scams, deep fakes, misinformation, or disinformation. Our approach, of course, is that we need to find the right balance. You want to have guardrails. But at the same time, you can’t have the guardrails being so tight that you circumscribe experimentation and innovation. This is a conversation that is happening globally. For us to be effective in this, we need to participate in those global conversations—which is exactly what Singapore has been doing. We’ve been talking to a variety of international partners to understand sensible rules for the road and ensure that AI can be used for the public good.
Let me give two examples. One was earlier this year at the World Economic Forum in January, when we announced the proposed model AI-governance framework for generative AI and sought views from the global community. That’s something built off an earlier model AI-governance framework that covered traditional AI. Another example is AI Verify. We launched an AI testing tool kit that helps organizations validate the performance of their AI systems against general AI governance principles. And, together with that, we also launched a foundation in 2023 that harnesses the expertise of the global open-source community to develop similar kinds of AI testing tools for the responsible use of AI.
Sanjna Parasrampuria: Kathryn, to complement the perspective that Terence has shared, are there any best practices from private companies’ perspectives on how they should think about their risk guardrails in general? And what are the trends you’re seeing as to how companies approach this?
Kathryn Kuhn: The first thing is that they’re asking the question about risk. Where before governance was focused more on a North Star architecture and compliance, now they’re expanding the notion of what it means to govern a technology. And in that case, they’re pulling in people who are not just well-versed in risk but also look at what the risk is to the company and the data, and what the potential implications are of that risk. Financial institutions are quite used to having compliance separate from risk but fairly mature organizations, such as retail companies, are not as well versed in those functions, and now they’re starting to build up that capability. You’re seeing this new notion of what’s the internal risk but also external risk, and how do I weave it in so that it both helps inform and guide your technology adoption but doesn’t stifle it; it doesn’t squash it down in the world of “no.” Nobody wants to be in charge of “no.” But, at the same time, you want to make sure that you guide people to ask what the range of risks is and what is it acceptable or unacceptable to us as an organization. That line will move. Risk is never really one and zero, it’s not a binary thing. It’s a range of possibilities. So, operating in that kind of ambiguity is also a new muscle for organizations.
The number of start-ups related to AI increases by the thousands every single week, so there are competitive pressures that may not have been there before. All of a sudden you have a competitor in your space and how do you respond to the same thing? There is such a broad range of topics that the private sector needs to respond to that they need those organizational muscles to be able to take in a lot of information and collectively make their responses.
Sanjna Parasrampuria: I’ve been involved with several clients on preparing them and thinking through what the workforce might look like. And while everybody understands that prompt engineering skills will be needed and thinks about how to train the data scientists or data engineers, the most interesting bit is a big demand for the role of AI risk officers and AI compliance officers. You don’t think about those naturally because your inclination goes toward thinking about the tech jobs, but these are certainly jobs that are becoming high in demand.
The last segment for us to touch upon is to think about how we advise businesses in general. Terence, I’d love to get your view on this because there’s certainly a very big opportunity here. How are you thinking about advising businesses that are wary of getting in? What is your advice to businesses today in Singapore from the generative-AI adoption perspective?
Terence Chia: Obviously, businesses are not homogeneous. There are many different kinds, many different sectors. At the risk of oversimplifying, you have big ones and you have small and medium-sized enterprises (SMEs). For big businesses, whether they’re local or global, they have the resources and the wherewithal to choose how they want to invest in generative AI and to upskill their workforce. And, of course, some of these things can be done in collaboration. A number of companies we work with, for example, on our TechSkills Accelerator, are among the bigger firms because they have that critical mass to be able to offer workers both on-the-job training as well as classroom training to help them level up.
My only request to them is, yes, indeed, you’re doing all this and you’re trying to ensure that your company will remain profitable and that you can benefit from this, but have a thought for the broader ecosystem and the workers. That makes it more sustainable in multiple senses of the word because it’s really not just about how you work toward your next quarterly report, but how you are going to generate growth, something that will last for the long term.
For the small and medium enterprises, however, the situation and constraints are quite different. The encouraging thing I would say to them, first of all, is that I think generative AI can give them superpowers as well. What is key, though, for these businesses is they need to understand how they can adopt AI or generative AI as a new tool in their tool kit. Here, IMDA has tried to lay a couple of things out for the businesses to make it clearer, easier, and more supportive for them.
At the national level, we are developing a Digital Enterprise Blueprint. The idea behind this is that it hopes to uplift enterprises and workers in this AI age by helping enterprises be smarter in using those solutions, helping them scale faster through the integration of digital solutions, and also helping them be safer with cybersecurity and resilience. That’s at a national level—but the magic then happens at the next two levels: the sector level and the company level.
At the sector level, every sector is different. You have the built environment, which is very different from the financial sector, which is very different from the infocomm sector, for instance. What IMDA does is work with sector leads from other agencies to develop industry digital plans (IDPs), and each of these IDPs has roadmaps for the enterprises highlighting specific solutions that the sector needs. We already have a bunch of these IDPs and now we’re incorporating AI-enabled solutions into them. We’ve done this for a few sectors already—retail, security, legal, and tourism—and we will continue to do it for others. We hope that by doing this, the SMEs (which employ two-thirds of Singapore’s workforce and contribute to nearly half of our GDP) will be able to effectively leverage the use of new technologies like generative AI.
This brings me to the last level, which is at the company level. The company may say, “Well, this is all well and good. You’ve got a blueprint at the national level. At the sector level, I can see that there are some plans and there are things I can adopt, but how does it apply to me as a company? And by the way, I’m a one-man show, or I don’t have enough people to hire a chief technology officer (CTO). So I’m really lost about this.” This is where we offer the CTO-as-a-Service. It is essentially a CTO in your pocket. As an SME, you can self-assess your digital readiness and needs and then get pointed toward solutions or even consultants that can give you an in-depth deep dive and help you with your digital transformation.
Sanjna Parasrampuria: There’s a framework in McKinsey that we use called “taker shaper maker,” where the takers are companies that will integrate off-the-shelf, generative AI solutions, which means that the cost of adoption will be far lower but the use cases far more generic. Then you have the shaper companies that will augment existing generative-AI models. They might take OpenAI API calls, but you typically see this in larger companies. The taker stance is taken by smaller companies where they don’t want to expend too much on creating the generative-AI technologies. And then you have some of the big ones, either in very regulated industries or information industries, where they take the maker approach and will build and develop their own machine learning models (MLMs). So that’s another construct that ties in with the maturity and the size of the company in general. Kathryn, anything to add on that?
Kathryn Kuhn: For me, the biggest advice for businesses is to take the long view, especially when it comes to your workforce. And the long view means two kinds of responses: one that is going to build your solutions and one that’s going to use your solutions. If we think about building up the workforce and closing that million-plus deficit of people who can actually build these solutions globally, the old ways of hiring and closing your gaps are not going to do it. The simple model of an onshore workforce and an offshore workforce also isn’t going to be able to work. So how do you think about closing those gaps?
And, when you think about closing those gaps, it means that you need to be able to consider a workforce that maybe isn’t the perfect fit because not everybody has been working in generative AI. So, how do you make the shift to a skills-based view of your workforce? And by skills, it’s how you find people who have enough coding experience of the right programming languages like Python, enough understanding of modern data technologies, such as R and Hadoop, and cloud proficiency. They may not know Amazon Web Services (AWS), but they know Azure, and therefore they’ve got enough relevant experience to make the leap to your technology. So, it may not be a perfect fit, but they have enough of the right skills that they can make the leap.
Underlying all of this, of course, is a proficiency in math. That proficiency is not only understanding calculus but also statistics, and really making sure that the core foundation is there. So, how do you make that shift to a skills-based workforce and close your hiring gaps in a new and different way? It also means considering pools of people that you might not normally consider. It may mean that you look internally and find people who you can easily cross-skill into the roles that you need. Therefore, you can keep those skills fresh so that, as technology continues to evolve, your workforce can also be future-ready and make the shifts in step changes.
Then, when it comes to the people who are going to use the technology, we’ve talked about critical thinking, but it’s also making sure that they come into roles that you wouldn’t have thought of as needing digital literacy, like a frontline worker. How do you make sure that that you’re testing for that core digital literacy, and again, that skills-based view of your workforce? So not just applying that to your software engineers, but broadening this notion of skills-based to your broader workforce.
Terence Chia: I want to reinforce Kathryn’s point about skills-based hiring. It’s something I strongly agree with and it is something that IMDA has been trying to support. We launched a skills-based hiring initiative late last year where we got companies to come on board and pledge to support skills-based hiring. To date, we have more than 200 companies in Singapore supporting this initiative. It’s very important because—both on the company end as well as on the worker end—it helps to maximize the potential of the limited resources that we have.
Sanjna Parasrampuria: I started this discussion by saying that 2023 was a big year for generative AI. It certainly was but it almost feels like now we are post-awareness but pre-deployment. As a data point, with all the hype around ChatGPT, it still has 200 million active users today. If we were to compare this with WhatsApp, which has been there for several years, it has 2.7 billion active users. So we still have a long way to go for the technology itself. How do we uncover this and what it could mean for our companies and workforces? Before I sign off, I have one question for both of you. In five words, is there any big thing you’re hoping to expect out of 2024 from generative AI?
Kathryn Kuhn: I’m looking forward to tempering, let’s call it, the froth or the irrational exuberance around this technology and instead start focusing on value. For every technology, there’s a sweet spot of value—capture that value instead of getting mesmerized by what generative AI can do and all the negative hype around it.
Sanjna Parasrampuria: “No hype, value capture” would have been your answer.
Kathryn Kuhn: Yes, ignore the hype, capture value.
Sanjna Parasrampuria: Terence?
Terence Chia: Partnerships. If there’s one thing that’s been very evident over the course of the last hour, it is that for all of us to thrive and make the best use of the superpowers that AI is giving, we can’t do it alone. Individuals can’t do it alone, companies can’t do it alone either. We need to work together, find ways to work constructively, not just from the perspective of governance, dealing with the risks and challenges, but also on finding a sustainable way to go forward and make sure that people, companies, countries, and really the whole world can benefit from what I think is a very promising technology.
Sanjna Parasrampuria: On that note, I thank you both very much, Terence and Kathryn, for taking the time. I’m sure our listeners are going to walk away with some very pragmatic takes on the technology: there is a lot of opportunity but there’s also a lot of work to be done ahead. And partnerships and value capture are the way to go. Thank you so much for your time.
