Futurists Gerd Leonhard & Anton Musgrave - The World By 2030, Episode 2 - Artificial Intelligence

Hi, I'm Gerd Leonhardt, futurist. I'm here in beautiful Cape Town, South Africa with my good friend and colleague Anton Musgrave, who runs Future World. We've done a few episodes on the world by 2030, it's getting more exciting and excited by the minute. We have a great topic today to talk about, artificial intelligence, the future of work and jobs. And it's not as dark as it sounds. I've said many times, I think, quoting, I think an old Supreme Court judge who said one machine can replace 1,000 ordinary workers, but no machine can replace an extraordinary worker. And that was 150 years ago. So let's talk about AI and all these things that are currently cooking. Let's start with ChatGPT, artificial intelligence, generative AI, and just as a quick start over this and then we'll talk more about this. There's been a great moment in AI and this has been going on really for 20 or 30 years, where the transformer model was started by Google. Deep learning, machine learning is what everybody's working on. But all of a sudden OpenAI decides to release ChatGPT. I think it was 3.0, now it's going to be 4.0 very soon to the public, more or less, to basically create a Sputnik moment. And I think OpenAI was fully aware of this, that it wouldn't work in most cases or it wouldn't really produce long term results, but they decided to jump the gun to create the sort of race to the moon. You could say that basically openais like Sputnik and Google and Microsoft are kind of trying to land on the moon. That's how I look at it. But what's your take on ChatGPT generative AI?

Well, I think, Gerd, firstly it's amusement at the paranoia, at the panic, at the amazement, if you like. I mean, is it interesting? Exciting? Absolutely. But let's not forget what happened in 1990, 1991, where we started connecting computers through a dial up modem and the world was amazed and scared. And Here we are 25 or so years later of fundamental change and we're only reaching the start line of, of that change. And ChatGPT and all its forms, other names if you like, is exactly the same. It's the start of something that's not going to go away. Is it scary? Absolutely. Is it exciting? Profoundly so.

Well, I think unlike the Metaverse, which is this vision of mostly meta in Zuckerberg, that we're all going to live wearing a helmet and take a Refuge in his empire. AI, I think, is much more of a long term lasting concept. I always say no, we have infotech, we have biotech, we have sustainable tech, energy tech, and then we have AI tech. It's of that magnitude. And I think what we're seeing with ChatGPT, to me is the pinnacle of that. I mean, basically just a tiny tip of the iceberg there, right? Because basically what it does, it regurgitates like an autocomplete, you could say, or some people say a fancy parrot. But it's also very much a dilettante. So it has actually no idea what it's talking about. And I think this is the best word for it is dilettanti. Somebody who looks like they know a lot and is very good at bringing it across, but ultimately has nothing underneath.

And there's no nuance, there's no perspective.

And there's no understanding. So I think this comes apart from the fact that this is a language model and LLM, a large language model. And a large language model uses language to find things. It doesn't use images, it doesn't use other things that are not data. It uses languages and it goes out to trillions of words and recombines them in different ways that make the most sense. And many times it's really amazing. So if you're having a problem connecting your camera to your computer and it will give you a whole tutorial and that's really powerful. As soon as you ask anything about opinion like should South Africa be dependent on coal or whatever, it will come up with something that sounds reasonable. But if you probe deeper, it will come up with complete conjecture. Because at some point somebody may have written somewhere around the world that South Africa is already out of coal, and they find that and they put it together. So to me, this is a very, very big question, is that this engine that produces roughly 3% of reality, we're looking at it as if it was reality. And that's what we always do.

It's like Lego blocks, right? I mean, it's like, you know, billions and billions and billions of Lego blocks words and it suddenly just finds different ways of constructing a sentence that reads logically in English? But where's the review function, where's the insight function, where's the nuance? And where's the connecting this set of ideas to an unrelated set of ideas in another industry and then drawing the joint inference from it that doesn't exist? But hey, gerd, it's only ChatGPT version 3.4 or version 4 soon. The point is, by 2030, it is going to be ubiquitous, it is going to be way more powerful, and it is going to take away many, many jobs. What we've got to figure out is how do we manage its impact and what impact we want it to have, firstly. And secondly, how do we then train humans, develop humans to have skills that a competent language processor can't replace?

Well, we've both talked about this for a long time, about routine, how routine is being replaced by machines. But a lot of times when we look at routine, it's only pieces of what we do that is routine. For example, you can outsource your bookkeeping to zero, and you can outsource your scheduling to slack or routines. But driving a car, for example, people think it's routine, but actually turns out it's not. It's very difficult to have level five automated driving. And if we get level three or four, we consider ourselves lucky and it's a great help. But we're still driving and we still have truckers. So we have this common misunderstanding that we're much further along and the machines will actually replace all of this. I believe that it will give us a chance to outsource the routine work, the monkey work, the commodity work, the dirty, dull and dangerous. You know, like in mining.

Right.

So that can be painful if you do, you know, if you work like a robot, a robot takes your job. But very few people work entirely like a robot. So, for example, check out the supermarket. If you work in a giant Safeway in America, so you have 25 cash registered and you don't do anything but just type and enter. Yes, job's gone. But if you work in a supermarket here in Cape Town and you have local relationships, you wouldn't want to take that person because that's the connector and you don't need to use a machine. So I think a lot of that will mean routines are being outsourced and by 2030, the system will be very smart. But also, we have to understand that in my view, logic alone is not enough. And these are logic machines.

Yeah, I, I think that's it. I mean, the fourth Industrial Revolution GERD was about automation, robotization, you know, technological everything. The fifth Industrial Revolution, I believe, needs to be about rehumanizing humanness and humans rehumanizing humans. So in other words, finding those features, those virtues, if you like, that are uniquely human, but that also add value to the customer experience in whatever industry you're in, those we need to enhance, elevate, monetize, differently and then let the machines do all the checkouts in a safe way in the US because you cannot tell me that that person is actually fulfilled as a human. You can't tell me they get goosebumps from what they do every single day. And very few of them make a real impact on the customer. So let the machine do that. But then we need to be courageous enough, curious enough, brave enough to say to humans, this is a way in which you can add relevance to society as a human, and then way to learn to monetize that differently. And so capitalism in a sense needs to shift, because we don't only measure profit on net profit, or EBITDA or market cap, et cetera.

This is one of our key challenges, clearly, because if we take them purely capitalist means, then you make more profit when you have a machine, because the machine doesn't have a union, it works forever, you make more profit. But there is a societal component. So when we're moving into a larger view of what we want, you know, people, planet, purpose, prosperity, not just prosperity, then this becomes a situation where say, okay, so we have all these machines doing commodity work. We have more productivity. What we have now is more productivity, but declining wages, you know, and declining jobs. And that's because of our policy, it's not because of technology.

Absolutely correct.

And so we need to fix that problem and say if we have inclining productivity, we have more money, then some of that money needs to go into a pool to redo work for the others. For example, I think we should have more healthcare workers, more social work, more people that get paid to have children, which is important. I think in a free society and all of these things, it has to be more equal. And that is the true problem behind AI, because AI will show us that routine work can be outsourced. Thus I only work two hours a day, and my boss will say two hours a day, I can get somebody to be four times two hours and I fire three people, right?

Yes. But I think what will also start shifting in the world is what we deem valuable. So an automated process will attract less and less of a premium as it becomes ultimately commoditized. So what then becomes valuable? It's those things that are scarce. In the world today, filled with smart machines, what becomes more scarce is empathy, insight, curiosity, innovation, critical thinking, etc. But the question is, how do we then monetize those differently? And I think the consequence will be that there are going to be millions of people that simply can't climb that ladder, perhaps. And so a Global universal minimum wage, whether it's that we tax the robots or whatever it is. But I think we're going to have to face up to that. Having said that, though, and we scared about the millions of jobs that AI is going to take away, but we haven't yet had the debate and explored how many more millions will be created in the industries you've spoken about, the care industries, for example, the career advisory industries and so forth, the personal strategy industries. And so brand new industries that will exist, that will fall a huge part of that gap, but not all of the gap.

Well, if the end of routine, as I call it, if that's not the end of work, then we have to shift our education and our planning into the things that are not routine. And that includes everything that's human. So negotiation, caretaking, complicated manual jobs, complicated intellectual jobs, but things with a judgment and on and on. And what we currently are doing is, for example, in India, we're producing 1.2 million engineers per year. And a lot of those engineers will end up doing the jobs that machines can do, which is how to figure out how to print a 3D house or how to lay a bridge across a river and then make $300 a month and the machines will take their job. And so we're barking up the wrong tree by competing with machines.

There's no question about commodity work. Absolutely correct. I think that's an absolute given. And I think that when you look at all of these jobs and what graduates are we producing for those jobs today in 2023, by 2030, we must have changed the entire global education system away from subject matter in grades. And every grade progresses at the speed of the slowest person in the grade. Those days are over. So the impact of AI in education, for example, as we merge education with AI, individual pupils will learn at their own natural rate, whether that's fast or slow. And you won't be one of 25 in a class that's graded simultaneously. But we're also then after going to give these new kids new skills. So the skills to think, for example, of a system, systems theory, critical thinking skills. How do you identify fake news, fake GPT output, or stupid GPT output from real?

Well, yeah, that's the other thing that's happening here, as the question is what is real and what is valuable? For example, you can say, you know you like Dall? E, so you buy a perfect replica that could be done by a fake Dall? E artist, or just print it. And you're perfectly happy with that because it looks great, but you know, it's not real. And the person that shells out $100 million for the real dolly knows the difference, but it looks the same.

And let that be the case if someone wants to pay part of the money. Part of the money.

And I think that is something that we have to also maintain by saying, okay, the real journalist is going to be more expensive because that work is real, it's not a copy. Right. And therein lies a very big difference. I think this is going to become apparent when we have a lot more AI writing books and stuff. They are like fast food. And some people are fine with fast food and some of it tastes okay, but you still know that it's probably not the best food you can eat and people have it and it's a choice. Right. But really what we want is real food. And I think this is where we're going to learn that the machines aren't capable of producing that because they're not capable of foresight. They don't have any emotions, they have no humanity, they have no consciousness, they have no spirituality. And basically the machine is as opposed to humans, which are all sensing, as people say. So when we speak, I hear, I see you, I smell you and everything. And this is what humans are. And a machine is only like one sense, and that is computing. They're binary. So I think we don't have to necessarily worry about the impact on that level, but also impact on media and democracy by those machines fabricating synthetic truth.

Yes.

Like saying, you know, look, when, when Britain joins the eu, all the Turkish people are coming because they're going to come to Europe and they're going to end up in London, you know, which was completely fabricated.

Absolutely.

Brexit. Reason behind the Brexit. Right.

I think another feature of AI is going to be that it removes lots of friction from the current global system. And if you look at the business models of many businesses, it's based on friction. Banking is a classic example. So it started with the Internet, that started accelerating speed and removing some friction. But there's still a black box in much of banking, asset management, currency trading, commodity trading, etc. But AI coupled with connectivity, coupled with speed of processing will start eliminating the friction. And of course, the moment you remove friction, the business model changes. And so companies are going to have to reimagine again what is valuable in the eyes of a customer.

Well, take a call center. Perfect example, classic example. So if you have a snowstorm in New York, airport is closed, 50,000 people have to be rescheduled. Most of them are not freaking top level flyers and stuff. So you have an AI, you call, you put in your PIN number or you put in your travel code, booking code and it will rebook you depending on what's the AI can do all of that. But if you happen to be making an unusual request or a change that's based on courtesy, the AI wouldn't know whether that's important or not. So I think what's going to happen to call changes is clearly 90% of that can be done by machines sooner or later. And so we're talking about roughly, I think the latest number is 18 million people. So that is definitely a job you do not want to take. But then there will be new call centers where you say we use the machine to get your data and then we have more fancy.

Well, I mean already there is, there's a young South African startup operating now across the world that use AI in call centers to map real time for regulatory compliance and emotional sentiment of the gore already in 2023. Now imagine this in 2030. But then again, there's a bank that I've had experience with that only employ graduates in call centers. They've used the technology to merge all their internal databases and knowledge systems. But the smart human can actually solve a problem across multiple departments. And for that bank you pay a premium and you're happy to do that.

Well, this is a question of general intelligence. Which humans are considered generally intelligent? Also because we have eight or 10, according to research, different kinds of intelligence, emotional, social, cultural, kinesthetic, you know, we have all of those intelligence. Machine only has one intelligence and that is computing. Right. So they are, they're binary. So and this is why having a machine that becomes generally intelligent in my view is probably not something that we should pursue as a top business goal. Because A, it does replace us when it gets to that point having an IQ of a trillion, say, right, A computing IQ and B, it would be almost impossible to control. So I always say we should probably be careful about how we use AI or what are called IA intelligent assistance now because of the automation and those things. And also make sure that we have a flag that says that it is from an AI. But we definitely are going to need a moratorium, a global collaboration agreement on general intelligence because we're not there yet at all because we don't have the computing power. But if we're going to have a machine that's generally intelligent like us, it will very likely be step and go through the steps of Evolution very quickly.

Elon Musk thinks we're living in a simulation already.

Right. So this is a very big issue. So we need to think about if we can collaborate across the world, and I think this is primarily Europe, US and China, of course, to figure out if we can control a general intelligence in such a way that it becomes a force for good.

Well, put it this way. I think if humans don't find a way of making the virtues of being human valuable, then we have a 20, maybe 30 year window, GERD, and then I think it's over for humanity. I mean, let's be honest, humans haven't covered ourselves in glory. Here we are tonight in 2023 and 3 billion people go to bed hungry. The system is not giving 9 or 10 billion people a fair hope that tomorrow will be better than today. And when we can't solve for that, and AI is one of those big issues. And of course the next question is, who do we trust to solve for these things? I wouldn't trust the un. I'm not sure I would trust the World Bank. You know, who do we trust with these big decisions? And how do we remove the vested interests, the Zuckerberg meta kind of influence over this conversation?

Yeah, and this is a very big question. I think that, I think that with technology like this, exponential technology, we need to always keep the human in the loop for the time being, even if it's slower or more cumbersome. So I think Microsoft has done the right thing with Bing by sort of castrating a little bit of its functionality and making that more manageable. Or it just says, no, I can't go any further.

But that's because of Satya Nadella and his humanness probably. Look how he turned that organization around. And it was based on culture and humanness and servant leadership. And so he has this, this benevolent patriarchal view over the power of the AI. But what if it's not him?

Remains to be seen if that keeps intact when there's trillions of dollars waiting on the other side of the inhuman behavior. But I do trust that they will find a way forward. But generally, I think keeping the human in the loop is to me is the ultimate for now, but also in the long run. Because the question of control is not right now. Control means, for example, that you can distinguish truth or non truth. And that needs to be regulated clearly in social media. I mean, imagine if you have an AI in social media that can Twitter every four seconds about the next president of South Africa and pollute the entire network of Fake things and videos and all the worries, good things.

I mean, let's not always see the negative side of it.

Yeah. But of course it doesn't sell. So the bad things go like six to eight times as fast. So that's why we have them. So having a story like the Brexit Turkey EU story goes fast because it's fear. And I think generally speaking, we have way too much fear about the future. When it's about AI, we should have caution and precaution to make sure that we can use it in the right way. But when we have fear about any of these possibilities, then we tend to demonize it or the reverse anthropomorphize. So we think, oh, chatgpt is my friend, like replica this app, or. Or this is my artificial friend. Or the AI should know better. When it's about criminal justice, well, it's.

Like a movie having a romance with its operating system.

Right, right. The movie horror. Right. And all these things. And we're a little bit polluted by what we see on television when it's about AI because the robots are coming, they take our work and then they harvest our bodies for energy and they take over. And we have to let go of that kind of idea and say, okay, we can use that for many, many things, but we're going to need a framework. For example, I am a strong proponent of saying we should have a flag. Every time you produce content that's done by the machine, you have to have.

A flag or have a human endorsement. And I think on that point, for me, often in these very complex and important conversations, gerd, we lose perspective on the essence of business. It's that moment. It's a moment when someone with a need meets someone with the ability to supply and they agree on terms and price. That moment is business. And what we need to say to ourselves is, how do we utilize all the technologies, including AI, to facilitate that? But how do we then close out that transaction with a moment of human magic? And only humans can do that. Even if it's all the processing power up to that moment, whether it's the handshake, the pat on the shoulder, the smile, or the human insight, the comment that really leaves a person with a sense of, I'm dealing with something, someone that I can relate to, identify with, and that can excite me about the transaction, that's the magic of business. Everything else is just part of the process.

I'm a little bit worried about this AI thing becoming like social media, where initially we were very excited about It. It was very magical, connecting with others, publishing your own stories, putting up your own images. And it became magic that we could do that and blog. And then it became regurgitation of other things. And then it became a game that we played, like, I'm the biggest Instagram, whatever, makeup girl, influencer, whatever. And then it became a toxic thing to where you would read all the wrong things and the people would kill themselves as to what they see.

I mean, I think Instagram and other of these platforms have a lot to answer for in terms of the social ills amongst young people.

But it went from, I always say, magic to manic to toxic. And the same could happen with AI. And that would be the end of life as we know it if we move from the magic part of this technology to manic, which means that we can't get out of bed without it. And then toxic would be polluting who we are and changing who we are. And this is a real issue also with democracy. How am I going to find the opinion when I ask my chatbot who to vote for? It would say, well, I know everything about you, and it would basically do the voting for me even without asking me. And I think that is removing our human authority of decision making.

And I think that is probably where we should wrap up the conversation, because it's that human authority. And I think from a business perspective, in the boardroom, leaders need to have enough knowledge to have a sensible conversation. It's 2030, AI hasn't died. I think we are on ChatGPT20 now, if I remember correctly. So the issue is, how do we equip ourselves with the knowledge to harness it, to use it, to manage it, to deliver the outcomes we want for our human customers, clients and colleagues? How do we allocate capital in the boardroom between technology, infrastructure? I mean, in 2023, there's cool banks that spend money maintaining COBOL mainframes. I mean, it's an insanity.

I always say we should invest as much money in humanity as we invest in technology, because that is the key to having a good future that we can make together. And the other one is, in terms of technology, too much of a good thing can be a very bad thing. And that's about food, water, drinking, eating, smoking, whatever, but also about technology. We can have too much technology, removing our authority over ourselves and getting lazy and abdicating.

Absolutely correct.

And this is the biggest danger by 2030. The danger is not that machines will take over, but that we become too much like them.

But maybe one of the Success criteria in your annual performance review in your company will be A, your ability to harness and deploy AI to make your own job more meaningful, and B, how have you shown your human impact in what you do as a worker? So you'll have two criteria. One is how do you leverage the tech? But then how have you shown yourself to be more valuable than the tech?

In other words, how have you moved beyond the routine. Exactly. And created new values on top? I mean, I think, for example, what's happening in medical field is a great example. Doctors will use AI and virtual reality and stuff to be faster, to be more informed, to instantly pull up information, to look at your melanoma and decide what it is. But in the end, the doctor can use the free time to see more patients or to become a better doctor or to talk to you. Right. Instead of rushing from one to the other.

I think if you look at the NHS at the moment, most people are going to see their doctors because, yes, they have an ailment, they have a sniff or a headache, but the actual precipitator for that was family tension for some other reason. The good doctor will understand that, counsel you and guide you around that, Solve that, and then the machine will give you a pull for the header.

Yeah, that's what I'm saying. I think a person with an AI will replace the person without an AI, but probably will not have an AI replace a person as a job, because all the other pieces are missing. A good doctor, I've met a few in my lifetime, they can look at you at 14 seconds and they've got an idea what goes on and how does that work? And so a bad doctor will just go and get you a pill because you have a headache or something. So I think this is really what's going on. And we have to figure out also how to protect humanity in this context and how to evolve our education.

But what you just said is the human with an ability to bring those human attributes to the forest in an engagement will always stay relevant and valuable. And let's not get too scared about that.

Yeah, and the other thing is, of course, to wrap up the conversation, I think by 2030 it's going to dawn on us that the future of life is not just all about work, like we had been discussing for, what, 30 years, that technology will result in us working less, as never worked out, but by 2030, I think we're going to have technology to that point where we can actually spend less time working on the commodity work.

Well, what's an enable look at what's happening today. There are these research projects all around the world where workers are working four days a week, getting paid for five days a week. And they found productivity impact, output increase, and they've as a consequence, been able to harness the technologies of today to create that space of freedom and companies now valuing the ability or the, the impact of having humans having more time to do what they want. The old work life balance debate. But productivity goes up, you see, it's not either or. It's not the old world of binary, black or white. We can have two outcomes in today's world, and that's the beauty of what the technology.

Yeah, I mean, 2030, 2040, we may well live in a world where work is no longer the central thing, but it's everything that we do in life and just work maybe one per hour a day for one person and, and for the other it could still be 15, but it's also much more about occupation and calling and especially I think, you know, looking at the people who do menial work and less sort of intellectual work, a lot of that work will be generated by all the things that are being freed up.

Gerd, there's a lovely story in Europe where a government is actually funding the education for young people on condition they live in a facility of old people and spend an hour a day with the old people. And you see, here's the economic model. The reduced cost on the state of healthcare for the old people when they engage with young people is greater than the education cost of the young people. So it still boils down to euros and cents, but it's a different way of thinking about the economic model. And suddenly the value of being just a human with an old person reading them a story is now translated into, into euros. So there's an economic model that works, but it's different. And I think we're going to explore by 2030 many new these economic models.

That's a great way to wrap up the session. Anton, thanks very much for tuning in. This is Anton Musgrave and Gert Leonhardt in Cape Town, South Africa with our show on the world by 2030. Thanks for tuning in, Sam.