LKY School of Public Policy: work, AI and the role of public policy

On Policy Unpacked, LKY School of Public Policy assistant professor Ruven argues the AI revolution did not happen overnight — 15 to 20 years ago, when he was building a data and AI practice at a consulting firm, a conglomerate founder stood up and said "we only want natural things, not artificial" and walked out. Today, AI is rewriting work. His own industry is the example: coding moved from insource to outsource to overseas centers under cost pressure, and is now coming back through AI — what he calls "misource," with each consultant paired with one or two AI agents working in parallel.

He frames policy in three legs. AI Strategy 2.0 keeps iterating, which he commends. The middle mile is enterprise SkillsFuture — each sector is at a different AI maturity. He proposes an AI Readiness Index, modeled on the earlier MCCI digital readiness index, so any CEO can locate their quadrant and get a tailored package. The last mile is policy communication. Only about 25% of Singaporeans have used their SkillsFuture credits, fewer than 10% used them fully, and the top three uses are flower arrangement, Korean, and car courses — a clear gap between policy intent and behavior. A Gallup study of 155,000 people across 140 countries shows 70% associate risk with danger and only 30% with opportunity — universal reskilling needs a mindset shift, not just skills.

The people who fall through the cracks are not the elderly — Singapore already runs decent programs for them. They are the non-traditional groups: full-time parents and the "sandwich" generation aged 35-49 who, because of delayed childbirth, are caring for young children and aging parents at the same time and have no bandwidth to reskill. Ruven points to Japan's micro-jobs platforms, Denmark's flexicurity model, and Scandinavian use of AI to help workers pivot to adjacent industries. He proposes the "honor nudge" — Mongolia gives medallions to families with four or more children, with no financial benefit, only honor — and argues this kind of horizontal trust and intergenerational bonding is the real catalyst for the "we first" social compact.

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

[music] Is your job safe from AI? Or is the real question, is your country ready? As artificial intelligence reshapes, Singapore faces a critical challenge. How do we equip citizens to thrive? And [music] how must public policy evolve to redefine the social contract between government, businesses, and workers? Welcome back to Policy Unpacked, where we unpack ideas, stories, and [music] people shaping the future. Today we're joined by Dr. Ruben, assistant professor at Lee Kuwanu School of Public Policy. [music] His work focuses on trust, social resilience, and the narratives shaping technology. Together, we'll unpack the future of [music] work in the age of AI and what Singapore must do to stay ahead. Welcome, Reuben. Thanks for joining us. >> Thank you, Suba. Great to be here. Great. Let's dive into the first one.

I'm quite excited to talk about AI today with you because PM Lawrence Wong at the recent rally speech also delved into this, right? So, let's get the ball rolling. >> What types of jobs are most vulnerable to AI disruption in Singapore? >> Thanks. That's a great question. So, like you correctly mentioned uh pres uh Prime Minister Lawrence Wong, right? He devoted a significant amount of time at the very start of his speech talking about AI. So that's not very common. Typically we are talking about setting broad agendas. But he dives straight into some of the forces shaping our society today. Now I think it's really important to understand that uh this AI artificial intelligence sort of movement is paradigm shifting. Everything is going to change because of this. But it's also important to understand that it didn't just happen overnight.

It was actually happening 10 20 years ago. Now you know two careers ago before I joined the Lee Konu school of public policy I helped a huge consulting firm start up their data and AI practice. This was about 15 20 years ago. So at that time um nobody knew what AI was. In fact we were preparing for months for a big presentation to a conglomerate telling them to think about implementing AI or artificial intelligence. So we prepared for months for this. Then we went uh there was a big room with the founder of the conglomerate there. U so we started talking. So I as a project director I stood up started talking about artificial intelligence. Within the first minute the guy just stood up and say that we don't want anything artificial. We just want things that are natural and then he just walked out of the room.

So all this months of preparation kind of went down the drain. But fast forward 101 15 years later what do we have in 2022 uh chat GBT came to public consciousness. I I I recall it was November 2022. And after that there's not one day that nobody talks about AI. So from, you know, then nobody knows what AI is. I I only want natural stuff. I don't want artificial. Fast track to almost everyone is talking about AI. It's almost an explosion of AI narratives, right? >> I mean, we're doing a we're using AI every day. >> Exactly. >> Some form of it. >> Exactly. Exactly. So, so I think uh the fundamental question is is then how will it change jobs, right? So, I think chop jobs will be changed in so many ways. I I'll just give you an example of the jobs I used to previously do, right?

So I used to be in consulting data strategy for governments and and companies and things like that. So during that time so that was about 10 15 years ago. Um coding was so important right but coding could because of cost pressures coding was starting to be outsourced to countries overseas centers of excellence overseas. So in the past we used to insource but it's a bit more expensive then we started to outsource right now I'm seeing us coming back full circle. So these days actually um this coding stuff um could be done at a more cost- effective manner using AI. So there's no need to outsource anymore. Right? So we are coming to insource. In fact we are moving towards from outsourcing to insourcing to misourcing. So essentially almost every uh consultant these days I speak to the big uh big firms they have what they call AI agents.

So for every consultant or or every employee there are one or two AI agents. busy. So it's all me sourcing now. So if I need to do stuff before I go to sleep, you know, I I I'll it used to be a different time zone, right? So we'll get them to do stuff. So I wake up in the morning, I I I get reports done, I get insights uh delivered to me. But now it's an AI agent doing that. So this is fundamentally disruptive. So what it means is that uh the jobs the task mix within jobs will change. So if the task mix within jobs will change which means that the people they are hiring to fill this jobs with a different task mix will be very different. So I think it it's quite fundamental.

Uh the other point I would mention is that some years ago I was doing a project for skills future Singapore and we were trying to understand how could we get Singaporeans and how can we push them towards the top echelons of leadership. How can we have a couple more Singaporean CEOs of Fortune 100 companies right? So I went on this mission to interview various CEOs to ask them about their insights. So one of the things that uh one of the views that really struck me was that um this this CEO was telling me that you know Singaporeans are very good at finding answers to questions. If I set them a question they oh I'm very good at finding answers but they're not so good at asking the question. But if you think about the AI revolution chat GBT as we are most familiar with right it's all about asking the question isn't it?

Without asking the question, AI can't work for us. >> That's right. So, I think over the prompts, right? >> Exactly. So, so I think there's going to be a fundamental shift of not just how to find answers to the questions, but how to ask the right questions. >> I think it's great that you mentioned uh you touch on skills future because that leads me to the next thing really. Um, you know, Singapore has skills future and now AI strategy 2. 0. Are these enough?

you know it's never enough because uh we are talking about a digital evolution uh rather a digital revolution right but we are talking about policy evolution so an evolution will never be able to catch up with revolution right so I think it will never be enough but we just need to try as much as possible not just to be reactive but also to be more proactive in our policy making scenario plannings and things like that so you know you correctly mentioned about the uh uh AI strategy 2. 0 0. So kudos to the Singapore government. Uh they came up with the AI strategy 2. 0 in 20 uh 23 and they have been iterative. So just before coming to talk with you, I I looked that the latest iteration was some some days ago, August 2025. So kudos to the Singapore government. It is very forward-looking and they always iterate uh um on the AI strategy.

But I think we we need to understand that policym AI strategy is but the first mile. We also need to think about things like what happens in the middle mile and what happens to the last mile because if the middle bar and the last mile stuff uh don't happen then the first mile won't get us to where our destination is. So I think the middle mile there are a couple of things that come to my mind. Now one of the the biggest employers in Singapore is really the uh small mediumsiz enterprises sector. So I think the middle mile uh really got to be focused on what we call enterprise skills future. Now uh the very early iterations of skills future gives uh certain skills credits to individuals to you know study what they want right pick up skills what they want but sometimes uh the skills are not too related uh to professional upskilling.

So the second iteration or the third iteration focus on enterprise uh sort of skills future credits. So I think that's the way to go because different sectors are at different maturity points when it comes to AI implementation or even AI appetite or what really works for them because at the end of the day we don't want it to be just implementing AI for the sake of implementing AI. It must be purposeful. So does it decrease your cost? Does it increase your your uh sort of revenue and things like that? So I think the middle need to focus on a very sectoral approach. So it could be down to first of all understanding AI readiness.

uh so there used to be this digital readiness index by then uh uh I would say MCCI or now MDGI so I think similarly I think it's good to create what we call AI readiness index so that any companies anymemes could come the CEO you know could do this readiness index to understand okay I'm actually in this quadrant and this is what I need to be moved to the next quadrant of AI implementation then uh the package that I get will be very different from say your company uh who is maybe AI mature and now want to be accelerated to the next level. So the package of things that you can implement and the skill set mix is very different from you compared to me. So this is essentially a more different strokes for different folks approach. So I think that's really something to think about. So meaning uh let's talk about what it actually means, right?

So, from the FnB industry, for example, I may be running a restaurant or cafe. You know, maybe I open at 8:30 a. m. , I close at 5:30 a. m. Now, uh, usually after lunch and before dinner, right, is is usually the time when nobody really comes. So, it could be an AI engine that takes into account the data that I have of people uh, you know, coming to visit and things like that. And then I can just ask the question to that AI uh intelligent chatbot to say should I open from 2:30 to 4:30 or what time should I open? So maybe I can I can save cost during those hours when nobody come but then there is potential in the evening right?

So so things like that I think uh AI is able to do but we need to sort of connect the dots for different sectors because they need uh kind of different things even for bigger companies for lift maintenance for example.

Um there were a lot of talks about how many lifts we I think we have maybe among public housing we have 28,000 to 30,000 lifts right and li breakdowns are very difficult right because of an aging population I need to go to the hospital right I'm on a wheelchair you know I'm mobility challenge and and things like that so keeping the lifts running is extremely important that's where AI can come in you know instead of being reactive you know let's say the lift breaks down I get the lift technician to go we can be a bit more proactive to understand what's the risk that this lift uh will break down and send the maintenance team before it actually breaks down. So kind of predictive maintenance, right?

So different um sectors will need different sorts of approaches for the society to get better because they they function right uh either for the economy or society. So that's what I mean by that middle mound uh different strokes for different folks from an enterprise enterprise perspective. Now the last mile is really to get all our citizens on board.

Now if you think of the previous uh skills future iteration um uh it was actually intended for professional upskilling but a couple of years ago I think there was some parliamentary questions about actually how many people use this credits right >> exactly >> I think only if I remember the stats correctly I think only one in four have actually used their skills credit and less than 10% actually used it fully but if you look at the top three causes that people use it for I think it was number one flower arrangement number two Korean number three maybe car right >> sure >> so there seems to be this mismatch of the policy intention and actually the policy implementation so I think what what really matters to close that gap is policy communications how can we use policy communications to nudge people uh to actually use their skills to credit and to use it in a way that will benefit them professionally so I think if we can think of strategy as the first mile uh different strokes for different folks as the middle mile and policy communications to really drive it towards the final mile.

I think we would be seeing great uh return investment of our good policy. >> So many things I'm picking up from you like the AI readiness index. >> That's really interesting. Maybe you should spearhead that from LKYSVP. Then the other thing is about um this last mile thing that you talk about the policy communication aspect. Um I remember when I was doing uh a course for public policy design and there was a policy problem that I was thinking about it was future and the exact thing was actually about how's the implementability really you know should something change. So thanks for giving me an idea actually >> I think we should spearhead this together. >> Yeah [laughter] well let's do this after the chat but uh but let's move on to the next thing.

Um I wanted to ask you should Singapore explore you know new policy ideas like universal basic um reskilling >> yeah that that's a very interesting idea uh universal basic reskilling the question is um what would those universal skills be right that that's always uh really challenging uh some thoughts come to mind I think the first is uh so so you know I'm really interested to do global studies and a couple of years ago we uh partnered Gallup to do the largest um uh sort of survey study of 140 countries about how people think about risk. We kind of need to know understand how people think about risk then we can communicate risk better to them. Case in point was the coid9 pandemic. Risk perceptions could influence whether they wear a mask or not and things like that. So what what we found is quite interesting.

So every time um you know when we ask um so this was about maybe 155,000 participants from 140 countries covering almost 99. 5% of the world's population. Now when we ask them what do they associate risk with 70% of the participants of the survey actually say that they associate risk with danger only 30% associated risk with uh opportunity. So I think therein lies an opportunity right we always think about uh some of these uh things as uh sort of danger so I don't want to do it um but not many people think about it as an opportunity so I think beyond just a universal basic skill is that universal mindset that we need to have that universal motivation that we need to to have uh to to sort of imbue in in people in Singapore and around the world that we yes we want to manage risk but we also want to see that opportunity.

So I think once we see that opportunity then we are able to think about what are some of the skills that can help us accelerate finding the opportunity but of course you know um when it comes to AI and things like that there will be increased risk of scams. So I think the other thing is to help people be a bit more vigilant to understand how do we balance uh not uh falling prey to scams but also seeing this AI uh sort of revolution as an opportunity. Then of course the other thought that I I have is uh back to how do we ask the questions. So it's not about finding the answers anymore because AI is very good at at doing that for us a lot better and a lot faster than maybe how we think we would usually get answers. So I think the fundamental um thinking behind it is how do we ask the questions.

So behind that uh is about deep thinking, deep work and I think that itself is very critical alongside uh universal skills mix >> and that speaks to critical inquiry and I think that that's how you need to have that emotion and the other thing I want to pick up was that when you mentioned fighting scams I mean you of course won that award for scam fighter right scam fighter award so it speaks to also the work that you do that sort of like encompasses this whole facet of the data AI and such. Um there's another question that I want to ask and let's move along.

Um inequality and digital exclusion who gets left behind >> actually um we tend to think um you know when I was reflecting on on this important question that you asked uh immediately my mind went to you know maybe older adults and things like that but actually those are demographic segments that we have very good initiatives to cater to. We have the uh digital initiatives for older people. We have very good programs in schools. So I I think I I want to take a step back to think about actually who falls through the cracks. Who are the unconventional because it's a revolution, right? So there'll be a people who fall through the cracks. So I think there are two groups of people who tend to fall through the cracks. Uh one is people who have left the labor market involuntarily or voluntarily.

So um so [clears throat] I have four kids at home so they're all very young. Uh so my wife uh made a sacrifice and and and left the job market to take care of our our kids right as they were growing up and we started having more kids. So I think for example full-time parents, full-time uh mothers, full-time fathers. Um those are probably one group that we don't want to uh sort of neglect. The other group is stepping away voluntarily uh from uh from work would be caregivers. Um and in this uh segment of society we uh psychologists call that established adulthood from 35 to 49. Um you know I'm in this group so I can speak for my for my age group. We face what we call career care crunch. Why is it a crunch period now? Because uh we are delaying child birth.

So it used to be we take care of kids when we in our 20s then when we are maybe late 40s or 50s or even 60s we take care of our parents. But now we take care of our kids into young kids 30s and early 40s and our parents are also had kids later. So what's happening is that in our late 40s early 50s we are also taking care of parents. So we have a career to to to to try to navigate but now we have sort of interlocking stages of life where we have to take care of kids and older adults and they all come together. You see so would I have time to reskill? would I have time to think about uh you know looking up for for whether my job is secure I won't. So even though I'm in the labor market I people like myself cannot really participate fully or or there's no spare bandwidth or time.

So I think we we probably need to um with this digital and AI revolution think about some of these groups of people who are who are left behind non-traditional groups because these are new groups that come to mind and think about how we can leverage um some of the different stages or interlocking stages in life to make sure they're not left behind. So for example, I know of many colleagues, brilliant data scientists, AI scientists who left their dogs because they want to take care of aging parents or their kids maybe before they leave um is is to tell them that look these are different areas to stay in touch with their skills. We probably want to take what we call what I call hyper local approaches. Now there's always a digital screen uh in lift the lending of every HTB flat right the vote deck.

So I think it's to bring on board I think as Prime Minister Lawrence Wong also mentioned in the recent national day rally speech uh it could be CDC LE right CDC or voucher fame if you remember them um but now maybe it's it's really to take a hyper local approach to be as inclusive as possible so that these unconventional groups which maybe we have not thought before would not be left behind. >> Right. Thanks for that.

I mean I I understand completely about the sandwich then generation that they talk about because I think a lot of my peers are also myself like that as well and I think it's a real uh big issue that faces a lot it's relatable um so let's see let's see we can actually take this further into the next question which is so what can Singapore learn from other countries like Denmark you know um Finland and South Korea >> so many things um you know these countries are breaming with good ideas they with good policy agendas, great um policy implementations. But I just want to preface to say that you know sometimes there are all these best practices, right? But you know best practices may not be best for us. So I think it's important to get some of these policy ideas and adapt it so that it works as best as uh possible in Singapore.

Now a couple of examples come to mind. I think we know that Singapore is on the cusps of being a super aged society which means that one in five persons in Singapore will be 65 years and above. So for that I think we can look towards Japan uh who has been quite brilliant in using some of these AI solutions uh to help keep uh older adults engaged. Now one of uh the things that they do very well is to have micro jobs. So it's so in interesting right because uh PM Lawrence Wong recently mentioned that one of the key initiatives uh I thought was very hyper local was jobs near you. So actually Japan has been doing all these jobs near you for a long time. They call it micro jobs but maybe micro jobs near you.

Uh so in wherever you are whether is it Miyazaki in Kyushu Saporo in Hokkaido and things like that there will be this micro jobs portal uh it's kind of online so you can use your mobile phone uh to maybe book uh a short work packet of 3 4 hours from 2 to 5:00 a. m. at 7-Eleven. So they make it really easy really seamless and it's sort of part of the culture. Now, it's really interesting because I was having a chat with my Japanese colleagues about how is it that AI is so well accepted in Japanese society. So, so they told me something I thought was quite interesting. They said, "Oh, because of manga, because manga, you know, uh all the time, right, you have all these AI robots or all different manifestations of AI and it's very well accepted.

So, people in their ' 90s, centinarians, they remember reading manga, which has been around for a long time. So it's it's almost as if it's already part of life from sort of fictional and and now it is actually happening in reality. And Japan is great because I remember when they had digital revolution 4. 0 they said that to to manage digital revolution 4. 0 we have society 5. 0. I think Japan is great uh to to learn from AI implementations for an aging society. The other thing that comes to mind I think is uh Denmark uh they have they they really help you manage transitions. There's this term called flex security.

So I think you know in this uh world of digital revolution there may be many job cycles will be a lifetime of different jobs or sometimes involuntary which means that there will be a lot more transitions compared to the past which is it might make the work uh or job market more dynamic but then it means a lot of relearning. So I think one thing we can uh sort of learn from them is how do we help workers manage that transition. Some of those transitions could be involuntary, no choice, right?

I'm retrenched like what's happening a lot now in the economy or it could be I really want to go to an adjacent job you know so how so AI solutions are abound one of them could be let's say for example um I may want to move from say my job currently in oil and gas to maybe the caregiving sector right the agent sector which is one of the fastest growing uh sectors of our economy so it might be interesting to first understand say this is my skill set mix uh and how is is related uh to the supply and demand of skills in the caregiving industry. Careg giving industry has changed in the last 5 10 years. Uh skills in demand in the caregiving industry 10 years ago may be quite different today. So maybe there are because it's not possible for me to pick up 10 skills, right?

But to understand what are the skills that highest in demand, lowest in supply will help me quickly pivot to solve another industry and AI can actually make that happen. So we run a lot of these models as well. How do we know? So how do we know from say uh job listings they're out there for 6 months and nobody's feeling it which means that there's a demand for a skill but in the supply we don't have. So things like this AI can actually help and the Scandinavian countries do that I thought very brilliantly. The third uh thing that comes to mind is is uh climate change in Singapore. So a lot of swings these days and some we wonder ah yeah such heavy rain again thunderstorm uh could we know about it the day before?

I don't know but I think in a lot of these uh countries uh US and and uh countries around the world they're very good at uh sort of weather now casting and forecasting and AI can help right with satellite imaging yes we I understand the challenges Singapore is really small and things like that but actually that could be a gamecher um to plan our day especially uh coming from a father of four kids >> hot cold not really it's hot and wet always here >> exactly So maybe let's let's look at this like looking ahead, you know, what should Singapore do next to rethink the social contract? >> So I think that the social contract is uh is a very is changing. Um so I really like what Prime Minister uh Lawrence Wong mentioned about having sort of a we first society.

Um so I think from a we first uh perspective what I interpret him as trying to achieve is that now because Singapore has been having a great uh trust dividend with the government and the people but if you think about the citizen and government that's kind of the vertical trust perspective right what about this horizontal trust between different uh diverse groups of people in Singapore that's where I think the we first tries to uh inculcate that sort of horizontal trust. So trust between uh uh sort of different groups of people. So I think that social contract uh we probably want to think about that. So with regard to that, it also means that we want to think about how people can help each other.

So there were a couple projects that I was doing for the banks in Singapore trying to get older adults you know uh to use more mobile banking, internet banking. So we always think about okay what can MEES do? What can the government do to try to communicate or notch? But actually when we did our research we found that actually among older adults um there are those who love tech so they will adopt it even before their grandchildren.

there are pe there are also older adults who are sort of no I don't want to do that right so they are lagging behind for good reason right they they don't want to do that I don't really trust it so what it means that you have the same age group of people uh some are very enthusiastic some are not so you can actually get them to help each other out we call that the cross uh activation of different segments so I think that is the way to actualize a weer society where we start to build trust uh laterally >> then the other thing that uh comes to mind I for social compact is that there's going to be a lot of social isolation uh because of digital and AI right so these days we hardly talk to each other right it's just >> me and my mobile phone >> so I think um to manage that social isolation what we want to think about is how can we have more intergenerational bonding >> so can we have more programs or policies that bring together grandchildren and grandparents so on one hand um we try to achieve uh sort of learning about AI together decreasing social isolation but also intergenerational uh sort of bonding.

So I'm always thinking about how can one policy or one program achieve several outcomes. >> Then the other thing is that um we I think it's quite important also to celebrate diversity. Now with this AI sort of revolution there are so many pathways to success. I can now almost build a system with agentic AI tools. So I it's very easy now to start a business. It can be a OPO, OMO, one one man operation or one person uh operation, right? Which means there are now so many possibilities. I think we want to start thinking about to sort of engender that social compact to feature divergent pathways and to celebrate them. >> So it can be as far-fetched as maybe somebody designing something uh you know for animals or for different interest groups. Uh it can also be uh you know parents with uh four kids or more. We want to celebrate that, right?

So, when I was in Mongolia a couple of years ago, I was just curious about how is it that the young people there one, you know, they like to have four or more kids and things like that. And I realized that it's because in Mongolia they give this medallion for people with four kids or more. >> So, as somebody who's very cheapkate, right? My next question is, hey, you know, uh, if you get a medallion, do you actually get 10% discount? And do you see? >> So, they apparently say, no, no, no, it's just for honor and glory, right? >> See, >> so I honor and glory and glory. Like I didn't think of that at all. >> Exactly. Because we're always thinking about what are the financial incentives to nudge action, right? >> But here I coined this as the honor nudge. Very little money involved.

But >> you you nudge people's behavior based on honor and recognition. >> I feel that we need to do a lot more of that. And that would actually be the catalyst uh to to build a more cohesive society. I think that could be the attitude and strategy that underlies our social compact. I really really enjoy the stuff that you you raised like AI readiness um like you know making sure that we don't leave people behind and how we actually look at policies to do that uh and and this bit exactly this bit about policy communications and make sure that we are kinder society despite and in spite of AI that that comes through so I really enjoyed this discussion >> thank you I thoroughly enjoyed it too >> thank you so much Dr. Ruven AI won't just replace jobs. It's redefining our expectations of government, employers, and society.

For Singapore, the challenge is not only to reskill workers, but to reimagine the social contract to ensure every citizen can thrive in a rapidly evolving smart nation. My thanks [music] to Dr. Ruben for his insights and to you for listening. Stay tuned for our next episode of Policy Unpacked.