You've probably heard some version of this headline over the last six months that A I or artificial intelligence is coming for our jobs that it might replace humans at the office. The World Economic Forum predicts that 14 million jobs will be lost over the next five years to tech. So how concerned should you really be? I'm Sona Ramesh from the Money Mind team. Here are five things to know about how A I is going to change the
way we work. And speaking today with Damien Joseph, Associate Dean from the Nanyang Business School and he lie founder of the A I startup E V dot A I, Jen. Why do you think that generative A I in particular has made such a big splash in just barely six months since it came into the spotlight?
I think it's because it's genuinely a very amazing capability that it's demonstrated, right? I think it's taken a lot of people by surprise, it's actually useful by the layman and it's kind of gone out from performing in the lab in other limited circumstances to actually being very useful in generating a
lot of interesting content for everyone. So I think it's actually crossed a threshold and I think it's taken a lot of people by surprise and it's a legitimate improvement in the state of the art for A I systems
prof you're already using chat G BT at work. And so are your students chat GP T came on the market sometime mid December. We are now at the end of the academic year and already in that six months, we have had students, right, using it for your final assessments, there have been other technologies on the market, right? That are able to produce lecture slides, lecture audios, right? Combining them together to
produce a whole lecture. I use chat GP T on a daily basis to summarize and get very quickly right into topics of research. For example, it gives you a very quick overview of a research topic, right? Let's say a literature review, great for that, but you need to understand its downsides, right? The downsides are that it can be inaccurate in certain places. It can give you references which are inaccurate, so it
takes a bit of effort. But as a general overview, if you want a quick summary of the things before you actually read the papers themselves, it's a very good way to start. Well, that brings me to my next question. How do we work with A I rather than against it? I like the question because it assumes that we already have to embrace G A I, right. And yes, we have to embrace because it's not going to go away. There are two ways by which we can embrace a G A I. One is to use it as a tool, right?
What I have been using so far by which the students also have been using as a tool to get something done. The other is to use A I in general as a collaborator. So as a collaborator, it requires certain skills, requires monitoring skills, evaluation skills and stuff like that. But it's also working with in many marketing situations. It's usually the first draft that is being written by A I and then it leaves the more senior experienced marketers
to polish up that work, right? So if you know how G A I S work, especially text based G A I S, right? It's basically the probability that the next word in a particular sentence is that word almost like predictive text, right? But that word that's gonna come up is being marshaled together by the whole corpus of text within its database, right? So at the end of the day, you get a mediocre type passage coming up, right? So that's something that most I think creative folks would like to get away
from Jen. What do you think is a I friend or foe at work?
If your job is to generate written content, you absolutely must start experimenting with generative A I models. So if you or writing marketing copy sales copy, if it's part of your job or you know, if you're an entrepreneur, this can really help jumpstart the content creation, right? So you definitely want to be experimenting, but you also want to
not take the output blindly. You want to be getting a feel for what it does well and what it doesn't, it's kind of like hiring this really super smart intern, but that's also weird and wacky. So depending on what you do, find it immediately useful in some ways to jump start your creativity, almost like an assistant. But I think it's still a watch and learn experiment and figure out what it is good for, for you for your role.
What about for more complex tasks? For example, I've heard of people using it to write code. For example, what's your take on that?
That's something that will need a lot more care. It's something that stack overflow, which is this site where a lot of people ask and answer coding questions has actually banned for now chat GP D content because it was generating code with errors. So when it comes to things without necessarily an absolutely right or wrong answer, marketing copy images, it's a great tool when it comes to things that have correctness, very
strict correctness criteria. That's when you want to be really careful, it might be helpful in generating something boiler plate, but you really want to use it very, very carefully. And depending on what you're doing it may or may not be worth the time at the same time. Experiment.
Lots of potential. Are there any shortfalls as well?
Yeah, there's two parts, right? One is to be aware today, when you are experimenting with these generative models, there are two things. One is the correctness of the output. The other one that I think especially larger enterprises are more concerned with is copyright and information leakage, right? So I think a number of large companies, Apple Verizon so on have forbidden the use of chat G BT or such models for work because that information might then be used for training and that
might end up being leaked outside with image generation. There's a question of, hey, is this a copyrighted Getty image that's been used as a base that was not licensed
like it or not A I is here to stay. So it's time we embrace using A I at work to complement what we already do. It can be used as a collaborator to generate content and to cut down on routine tasks. But it's still very much a work in progress with humans needing to check for inaccuracies and to fine tune what the A I produces. Another shortfall is copyright and information leakage with some companies actually concerned about the data being fed into training models that could be leaked.
So several reports have suggested that jobs will be lost due to the rise of A I and tech like that W E F report I mentioned earlier that predicts as many as 14 million jobs gone over the next five years. Do you agree that jobs will be lost? Or that companies might hire less? It may hire less or they may hire more? Because one of the other characteristics that these tools or collaborators work faster, work more efficiently, they are more productive, they
are more accurate. So the throughput is faster. So it then adds pressure on the upper layer to take in all this work from the lower layer and to use it. If you have too few people at that layer, there's going to be a backlog. So there will be a pressure hire more people right at the upper higher value added layer. So the thousands of jobs that will become displaced, that could be at the lower level jobs. So yes. So in these kinds of situations, we have to res skill,
the folks we may have to upskill them. But can these people then move up to higher level jobs? Probable, not everybody has a capacity to, but I'm a firm believer that the majority can move up.
The reality is there's so much work to be done and it has always shifted as we've been able to automate when ATM S started widely deployed, people predicted you're gonna lose all your teller jobs. But actually because it made the cost of opening branches much lower, the banks actually open more branches at one point. Right. So it's unpredictable but the role of the teller change from, hey,
you're not counting money anymore. That's what's what the ATM does, but it's tackling the harder problems, things that need more judgment, more complexity, uh and building relationships with, with customers. So I think the nature and the role shifts but not the actual amount of work that needs to be done. That's I think what people need to adapt to. You're not wedded to a particular role, you're there to create value.
And I think you, you create value through, of course, seeing what needs to be done doing that, you show up, be responsible, those basic rules don't change, right? And at the end of the day, it's about people, it's about relationships and the, the A I Automation systems are really there to free us to focus on the things that are really important. So there will be, of course, structural areas where you're going to find that maybe if I needed 10 people previously, maybe I need 3 to 4, right?
Uh And maybe even one but large companies to do a lot of this employment, they don't necessarily do this very, very quickly. So a lot of times it's just by attrition freezing hiring rather than firing people. I think overall, you, you're just going to see the nature of roles shift as it has been shifting for 100 years, but you're not going to get mass unemployment per se. So what sort
of skills will people need to acquire to keep up with all these changes? Some of the skills that now we need in working with A I are evaluation and monitoring skills, right? One thing about A I is its ability to learn, meaning that it also will make mistakes. So we need people to really be vigilant about the accuracy about evaluating the outcomes of any A I and training. So then now the worker becomes, is also some form of an educator and a trainer
of that particular system. So that's another whole set of skills that you have to get. So will everybody do it? No, my research shows that, you know, some people will actually take on that role to train, but others may sabotage the system because why they fear that this particular A I tool is going to take away their work. But really not with proper upskilling with training provided by the organization, you get to go up to those higher value type jobs like monitoring, evaluation, supervisory.
And so that's where I think the workforce will go.
And also people management, your role becomes more about dealing with people about understanding problems, looking at how you can do things better, which are in general skills that have always been valued anyway. So you're basically being a manager, you're being a manager of A I systems and then managing upwards as a whole, we're bounding this a lot where knowledge work is concerned. A lot of the trades are ironically not affected at all. Right.
You still need a plumber, you still need an electrician. You know, these are highly skilled jobs in their own, right. And, and they are less likely to be disrupted in the, in the short term, right, or supplemented. So we are at the end of the day talking about knowledge work which tends to be bound in, you know, corporate kind of scenarios. Yeah. And you know, I think you're looking at maybe 30% of work evolving and being restructured in some way.
Prof Damen, tell me a little bit about this A I divide that you see potentially happening. This A I divide. It is not so much about the types of jobs that will be displaced and all that, but it's more lower level, it's about performance. So I'll give you an example of our students, students who have access to chat GP T 3.5, produce work in a certain way. Students who have access to chat GP T four by paying $20 a month for subscription,
produce slightly better outputs. So here we already have a divide based on the resources that one is willing to expand. So think about that in terms of performance at work, people who have the technologies who have access to the technologies, people who have been trained in those technologies versus those who have not, it's gonna widen even in that same job role. So that divide is more than just job roles. It's about performance, it's about rewards to their performance. It's about promotion.
It's about careers. That's how micro our digital divide will become. So upskilling is really the name of the game. Are companies ready to upskill the scale that's needed. Yes and no. If you look at Singapore's training participation rate, it's mixed according to the level of jobs that we find in our labor force, the professionals are retraining and training at about a 66%
rate in terms of participation. So that's good. So they are keeping up to speed with whatever new technology, new work process out there. At the other end, we have the lower level folks, the clerical folks, the production folks who are doing it at less than 20%. So now we have a divide where the professionals are upskilling with these new ways of working, the new tools are working but the people who are doing the most
routine work, they are not being trained as much. So really then it is now a need to go in for organizations to figure out what are the obstacles of these folks from going for training for res skiing? It's just not a matter of saying that yes, we can go for training, right? Are they able to do so? Are there support structures out there? Right? For them to train? Are they given leave if they leave, who else is going to cover their work and cover it efficiently to
the best of their abilities, that sort of thing. So there's a lot of discussion happening. So yes and no, there you have it A I is here to stay. But rather than see it as a competitor at work, we can use A I to complement what we already do, sort of like a supercharged personal assistant. Now, while it can help with tasks like content generation or bookkeeping, more complex tasks are still a work in progress. There are shortfalls with inaccuracy being the big
issue. So the onus is still very much on us on humans to spot errors and manage these A I systems at work. While certain roles might be at risk as we use A I, more and more experts say that people don't need to fear mass layoffs. Instead roles will shift to become more about people management and higher value tasks working with data that's churned out by the A I for example.
And to keep up with all these changes, lifelong learning and upskilling is key with companies, governments and employees all having to buy into this culture of learning to keep up. If not, there could be an A I divide where those who know how to use A I to do better work sprint far ahead from those who don't. That's five things you need to know about A I
and the future of work. My thanks to my guest, Damien Joseph, associate Dean from the Nanyang Business School and Gin Lee, founder of E V dot A I Money Mind is every Saturday on Media Corp C N A. You can also catch us online at C N A dot Asia and on youtube.