Singapore's view of AI has shifted: Josephine Teo interview

Josephine Teo used IKEA as an analogy for how AI adoption spreads. Individuals have hit their "IKEA moment" through tools like ChatGPT — they try, fumble, and improve. But enterprises haven't. Singapore has built around 65 AI Centres of Excellence; the first ten were hard, the next ten easier, and by the 30th the pace surprised her. The target is thousands. For that, testing and governance frameworks are not built for end users to verify themselves — they target the "aggregators," the IKEA-equivalent institutions that bundle AI tools with assurance before distributing them.

Frontier AI companies choosing where to invest in Asia weigh several things: a market open to new ideas, access to networks of like-minded talent, and partners with ambition who can grow alongside them. Singapore has the bench strength, growing appetite for risk-taking, and visible government investment in R&D and science — not just data-centre hardware — that gives frontier firms reason to anchor here.

Singapore's talent framework has matured. Two years ago it broke talent into AI creators, practitioners and users. Now it's an ecosystem view, modelled on Changi airport: each terminal must be excellent, but the real value comes from connecting them — moving people, vehicles, baggage, even horticulture across T1–T5. AI's real impact won't just be each industry transforming alone; it will come from manufacturing, healthcare and banking interacting in new ways. GovTech's platform already lets public officers build their own AI bots — over 30,000 exist. Most aren't great, but enough are to shift how work gets done.

Caption language: en · Fetched: 2026-05-02

What makes Asian and Singapore a good place to invest when it comes to AI and a bit the future of Agentic and some of the key tech moves the ecosystem players are going to make. Well, you know, the landscape is a very wide one and um when companies look at what they would like to devote their time and attention to doing um it also depends on you know what their starting points are. So if we just look at one type of companies, companies that we could describe as being at the frontier, um when you are a frontier company, um and you're thinking about how best to deploy your resources, um the idea that in Southeast Asia there is a market and this is a market that is receptive and at some point in time this market is interested to invest. This market is interested to try new things. I think that is an attractive one.

Ying already spoke about that. The report also covers that quite extensively. But then when they go to the second level of decision making the conditions that they are looking for that I think um throws a bigger challenge to them because uh it's one thing to find a location where they can access compute they have basic infrastructure that they don't have to worry about. uh it's quite another thing whether they can access a pool of talents.

uh as companies you know when they relocate from one place to another or they start up a new place your first series of actions almost always involves poaching right so you must have a you must have a rich environment and rich ecosystem from which to poach in order to get started it's not necessarily a bad thing it creates a certain dynamism and vibrancy in the market uh but if this ecosystem you know of talent isn't expanding then you know we're just poaching from one another it doesn't grow. So it has to be a talent environment that is also organically growing because people are attracted and why are they attracted? Maybe it's because you know the living environment is not bad.

uh but I think when it comes to AI talents who are sought after everywhere in the world um their interest is in plugging in to networks you know where there are like-minded talents people who are engaged in solving uh interesting challenges people who bring something to the table they have a new insight they read a paper um they came across uh this other AI researcher or they came across someone with domain expertise and they are trying out this you know very interesting idea and you kind of need to have that buzz to plug into that's a very important consideration but I would say that for a company at the frontier of AI probably the single most uh important factor that they consider is who can grow together with me because I'm devoted devoting all of my precious time and resources to try and identify a worthy use case.

The organization that owns the use case must not only have, you know, the resources, they must have the ambition, they must dream big, they must have the wherewithal to grow along with me. And if I can only pick a few partners to grow together with, where can I find them? And the the the reason why this is so important is because when you are at the frontier, it's so competitive. What you lose is time and opportunity. If in a way you invested it in with the wrong partner, the partner didn't lead you anywhere because of whatever reasons, you know, your experimentation just didn't go far enough. So I think this what's this is what makes Singapore interesting. This is where you will plug into a healthy ecosystem. I think the bench strength is here. Uh but what I also see is that there is growing appetite for risk takingaking.

There is a certain amount of dare. There is uh imagination and when I spoke with a number of um you know the leaders of frontier AI companies, that's what's drawing them. That's what they're thinking about when they, you know, look at the landscape and they say, "Here is this interesting market. How do we, you know, find ourselves a gateway? How do we find ourselves a foothold? " I think there is just one other thing I would add. Um, these individuals, these organizations when they're doing their calculations equally, they look at where you are putting your resources. It's not harmful if you are putting in resources to build up data centers, you know, uh but that's really hygiene. That's very baseline. They're looking at whether you are putting in resources into R&D. Are you investing in science?

uh if you are doing those things then perhaps they see that there is a greater uh headroom for them to also uh bring what they have that is the best that can contribute. So I would say that these are the number of factors that u you know help frontier AI companies think about Singapore when when they try to choose a location to access the Southeast Asian market as well as of course you know the um uh the broader Asian uh market. The two words that come up a lot are speed and trust. And so you know this is of course a technology that moves a lot faster than regulation.

And uh and you know what role could Singapore play for more responsible if you want experimentation and also for unlocking speed because to some extent some people are a bit constrained and there might be fear in the system or too much risk but uh what role could Singapore and Azan play where people could actually lean forward and go faster while maintaining the trust with the technology? Daniel, I guess when you think about speed, you you're thinking about, you know, how quickly can we scale the adoption to the wider economy so that it doesn't stay only within frontier companies. You want to find a way for it to diffuse. Um, you know, I think at the individual level, and the numbers bear them out, at the individual level, many of us have experienced what I could I could refer to as an ika moment.

ana moment in the sense that well when you use chat GBT you realize that it's not that difficult to use when you experiment with notebook LLM and then you realize that oh you can create a podcast too it's not that difficult if you use uh nano banana you know okay you can create all these fancy things right and then that is the ia moment for us at the individual level but if we unpack it a little bit how does this IA moment come about um this IA moment comes about well using IKEA as an example. Well, first time you step into IKEA, you've never assembled anything on your own. But you see that there are many people you know seemingly um very confident about buying stuff that they can bring home and then you know uh build up. Um and so maybe you think that I can't be all that bad. Let me try it too.

Um then when you bring it home and you unpack this package, what actually helps you to get the job done? There has to be some simple instructions. The way the object, the product is designed is relatively easy for you to assemble. And maybe the first time you build a bookcase or a wardrobe, oh no, maybe not a wardrobe for a start. a bookcase and maybe assemble a some simple tables. You know, the three legs don't exactly balance and it's not perfect, right? But you gain some confidence and the next time you buy an object, you assemble it slightly better. You you you pay more attention to the the instructions that are given to you. The ika moment for individuals came about with chat GPT. You see other people using it. that find that it's not that difficult for youth to use it. But the e moment for enterprises, that's still not far enough.

It's not gone wide enough. Um, we have maybe about 65 AI centers of excellence built up already. When we first started, we kind of thought, who would be so brave as to think that they could get this off the ground in Singapore? And then we saw something interesting happen. The first 10 was not easy. The next 10 was a little bit easier. By the time we got to 30, they were updating me. Eing team, they were updating me at such a fast clip that I was pleasantly surprised. And it seems to me that at least among some of the leading companies in Singapore, the IKEA moment has arrived. >> Yeah. >> But we are not satisfied with 60 hundred AI centers of excellence. This is a democratizing technology. Rightfully it should get to the thousands, the tens of thousands, right? And that's something that we should work towards.

Now then to your question, how do you get to this kind of scale and how does experimentation come about? It's not going to be the individual buyer of the Iaya products that do the testing. You buy it knowing that it was built with a certain degree of reliability. Um, in the early days, I think they they made a great effort of showing you how many times the sofa was jumped on, you know, and and and and the the the the beds can withstand, you know, your kids uh, you know, playing a fool on it for for for I don't know how many times. But the same concept applies when we invest in building up our testing capabilities, when we develop our governance frameworks. These are not necessarily intended for the last user of AI to go and you know satisfy themselves with. It's going to be the aggregators.

It's going to be the equivalent of the IKEA that has to invest in this effort so that when they distribute the AI tools it comes with a certain assurance. It comes with a certain seal. It comes with a certain certification of reliability and that is the journey that we are going through right now. And I'm I'm reasonably uh encouraged by you know the response. Uh certainly I think within the ecosystem in Singapore there is a good understanding that if you want this technology to go further faster you need to assure the people who are using it that it is safe. It's a little bit you've heard this many times before you can drive faster you can I don't know that you actually shift to fifth gear now every just step on it but only because we know there are seat belts there are airbags and you just have to develop these systems.

So this responsible experimentation we have to think about where it needs to happen. It's not going to be um at the individual level or the enterprise level. Very often it has to be at the level of the aggregate that uh gets this done. >> One of the things we picked up as you work with the private sector is last year was all about use cases.

This year it's all about what's the reimagine moment and how can you reimagine your company your country agentic first if we were and I know you have always bold ambitions if we were to go out perhaps two or four years and you think about what you would have achieved what are some of the markers for Singapore in terms of leading leading the way what's that reimagined marker for Singapore and this AI revolution [snorts] >> I think it has to be in you know the people that you speak with how they are describing their work to you >> today uh compared to four years down the road. It has to be the business owners and how they're describing to you their next big idea, how they're going to grow their market and how they're finding new and interesting ways.

And I think when you speak with individuals, when you speak with enterprises also, if you speak with government officials, you see that uh there is a change in the way they think about their work. Yesterday I was um was it yesterday or maybe the day before um our government technology agency, they moved into new premise in um Pongo digital district and so they had a little you know sort of among stuff staff only opening. So I was there and um our team that uh is is uh helping uh the whole of the public sector to think about the use of AI uh they told me that the platform that they built it is a platform that enables public officers to build bots AI bots for themselves. So for example Daniel you are you know a colleague who is so inquisitive.

You're always asking me questions and in answering all of your questions, I have no time to do my own work. And imagine if I have 10 colleagues like you, you know, constantly doing that. What if I build a bot to answer your questions? That's exactly what many officers have done. >> Good idea. >> We have something like 30 over thousand bots built amongst public officers. Are all of them fantastic bots? Of course not. But a good number of them are actually pretty good. >> Yeah. And this is the kind of momentum that we'd like to see. The enthusiasm, the aha moment has reached the individual. The aha moment has reached what they can do in the workplace. It needs to reach the whole organization. It needs to, you know, uh bring about a certain excitement across different industries, sectors and then you grow it like that. So >> amazing.

Well, you know, once you have that bot, perhaps we can share the bots. That would be very very helpful. Yes. >> Quite easy to build. >> Exactly. Exactly. Uh I know we're going to run out of time, but I think we had one question that we also sourced from the um from the from the audience and you talked about talent uh in in your in your first remarks. Um you know when we work with clients it's very clear that any AI transformation is actually it's not technology it's actually a people transformation and um how you had a lot of ideas on how do you make Singapore a bit of a hub for a AI enabled workforce. say a bit more about a bit this vision you have on on on how you actually solve the talent side of AI and make Singapore and Azan a hub.

>> Um in just the two years since we refreshed our national AI strategy, I think our thinking has also shifted. The way we look at talent has also shifted. At the at that point, we could kind of think of talent at three tiers. We were thinking of AI creators, you know, people who uh design the most sophisticated models and they put out AI tools that uh uh the world wants to use. Um then we were thinking of AI practitioners. These are basically data scientists, machine learning expert um you know engineers who will have to then bring these ideas back into their specific context in in order to make sure that there is an actual useful application.

uh but we also thought that at that point in time if you didn't have the broad base of AI users right it couldn't you know support um uh AI innovation in a very uh significant way so a very simplified way of thinking about AI creators AI practitioners you know and AI users but I think our our perspectives have evolved uh the AI talent ecosystem is much richer than that and um the best example I could give to you is perhaps uh when you think about our airport. If you think about our airport, you think about how you get a terminal going. Somebody has to design the terminal. Uh somebody has to get it built. After getting it built, someone has to make sure that there are systems and processes to maintain it on a regular basis, right? And someone has to think about what are the most useful technologies to apply to make this terminal run well.

Now comes along somebody else and says ah we have this new exciting great idea right and then they have to also assemble a cast of talents to build yet another terminal. But if you ask yourselves even just Chinese terminal uh Chinese airport T1, T2, T3, T4, in future T5, what makes this airport so outstandingly, you know, great? Of course, each terminal has to be great. But you don't as a user, as a traveler, you know, stick with one terminal. It depends on how the airlines have routed and it depends on how we optimize the slots that are available to us. So when you think about building a new terminal like T5 on a standalone basis, it is outstanding, but its real value is unleashed when it's able to connect back to the existing, you know, cluster of terminals.

And to do that is really difficult because you'll have to think in terms of well how do the people move from this terminal to the other terminals? Even if you can move the people, how do you move the vehicles that move them? And very importantly, how do you move the baggage? So that requires a whole new way of thinking. Now then if you think about this is just the hardware, the infrastructure, uh the logistics of getting things moving around, you don't just want a shell where people travel in and out of. You want it to be an experience. Then you think about, oh, you know, the other terminals, they are very green. Can we bring the same type of greenery? And then you ask yourself, what makes it possible for you to deliver it? Well, you need your own horicultural experts.

I don't know how many airports hire horicultural experts, but Changi does. Now, if you extrapolate this and think about what AI could potentially do for us, AI transforming manufacturing, AI transforming healthcare, AI transforming banking, the real value is not just that they are individually transformed as industries, as enterprises within the industries. It's when you add them all up and they interact with one another in brand new ways that were not required before but also not as impactful as before. So in a way when we think about talent the the the the the need the ambition also has to grow. Yeah. You can't just think about them in a very narrow you know way. You have to think about the whole ecosystem coming together with talents in every nook and cranny at every level and and that's what I think will will give us the most returns.

I don't want to give too much away. In any case, we run out of time. Um, as it turns out, in less than 30 hours, uh, the prime minister will deliver, uh, who is also finance minister, he will deliver the budget statement and he will have more to say about it. But what he says, what he's going to say reflects this sort of maturing of our thinking about what it takes to make this technology really come alive and enable this democratizing generalpurpose technology to really give us an uplift um that we have not even possibly imagined the full extent possible. Well, Minister, uh, thank you for this exciting vision about about Singapore.

By the way, I love that point of this multiplier effect of talent like your analyia on the terminals, but thank you very much for this discussion and and I think we're out of time, so we're going to we're going to wrap, but perhaps a round of applause for the the minister. >> Thank you, Daniel. [applause]