Will AI drive explosive economic growth or is it just hype? | Counting the Cost

Hello, I'm Adrian Finnegan, and this is Counting the Cost on Al Jazeera, your weekly look at the world of business and economics. This week, it's seen as a game changer, but critics say artificial intelligence is just hype. So will AI drive explosive growth, or will it disrupt economies and cost workers their jobs? AI technology is costly, but tech companies are pouring billions of dollars into it. Will the investment pay off? And just how important will AI be to a business's bottom line? And the UN says the need for global regulation of AI is irrefutable. It warns that if the industry is left ungoverned, its benefits could be limited to a handful of states and companies. Almost $1 trillion. That's how much tech companies are estimated to be spending on building up the artificial intelligence industry over the coming years. Supporters of the technology say that AI will increase productivity, boost incomes, and revolutionize the global economy. But with the excitement surrounding the potential of artificial intelligence comes alarm. A growing number of investors worry that AI won't deliver the huge profits they seek, at least not in in the foreseeable future. And critics warn that AI could replace human jobs and deepen inequality. Calls are growing for regulators to come up with policies to leverage the potential of AI for the benefit of all humanity. Fintan Monaghan reports.

The hype around artificial intelligence has created high expectations. JPMorgan predicts that it will eventually add as much as 17% to the US economy. And many industry insiders have hailed it as a revolution that will change the world forever. It has huge transformative potential, like akin to what the steam engine did for physical labor. You know, it took loads off the backs of humans, off the backs of the animals that we used to have plow our fields, and put it onto machinery. I think that's a good analogy to what these large language models are going to do to intellectual labor. But so far, that change hasn't quite materialized. Businesses have had mixed results. Some case studies have shown productivity gains, while others dropped their AI services altogether, saying it wasn't worth the cost. Some critics have called it a bubble. They question whether generative AI is currently up to performing complex tasks.

They're good at doing certain things very well, like translation, sorting, um, They are very good at pattern recognition of human faces. Humans are still infinitely better at that. I think it's incredibly overhyped.

But talk from skeptics hasn't stopped huge amounts of money from flowing into AI. Venture capitalists have invested more than $60 billion this year. On top of that, big tech firms like Amazon, Microsoft, Alphabet, and Meta are each estimated to have spent between $30 and $60 billion. Analysts estimate that returns are still dwarfed by expenditures, but firms believe they must secure a slice of the market or be left behind.

When you go through a curve like this, you know, the risk of underinvesting is dramatically greater than the risk of overinvesting for us here.

In the short term, AI faces obstacles and opportunities. Costs to the consumer appear to be going down. The latest models from Google and OpenAI are both faster and cheaper than their predecessors. But governments across the world are also debating how to regulate AI, and it's unclear what impact these regulations will have on business. Some analysts say AI won't have a measurable impact on national economies until 2027. So perhaps time is what will make the difference and reveal if AI is boom or bust. Fintan Monaghan, Al Jazeera, for Counting the Cost.

Well, companies and governments around the world are weighing up the costs and benefits of artificial intelligence, integrating AI into a company's business systems isn't cheap. Research firm Gartner estimates the cost to be between $5 and $20 million. It also predicts that 30% of these projects could be abandoned by 2025 due to the expense. But there might also be much to gain. Consultancy firm PwC estimates that AI could add nearly $16 trillion to the global economy by 2030. Well, with big changes come big disruptions. The IMF says that 40% of all jobs will be affected by AI. In some cases, it'll make those jobs much more productive. In others, though, it could displace workers as many tasks previously done by people are taken over by computer programs. All these changes won't impact everywhere and everyone in the same way. Developing economies are likely to face less disruption from AI compared to developed ones, but they may also miss out on a lot of the gains, potentially widening global wealth inequality. Well, South Korea is experimenting with AI in one of its busiest subway stations with foreign tourists in the capital. The new service can interpret several different languages in near real time using the technology. Eunice Kim reports from Seoul.

We're at Myeongdong subway station here in Seoul where a new AI-enabled service window has been installed. It's able to facilitate simple exchanges in 12 different languages, excluding Korean. Let's go check it out. Hello, where is the nearest bathroom from here? The system translates the speech to text in near time on the screen.

If you look to the right, you'll find the bathrooms.

The stationmaster provides an answer in Korean, which is translated into the inquirer's language. Myeongdong Station was selected as a testbed as it's most frequented by foreign passengers of all metro stations, says Choi.

When approached by a foreign visitor in the past, I'd become frozen with fear because I don't know how to speak their language. Now I have confidence that I can provide friendly and accurate information thanks to this new system. I'm very pleased with that.

The AI feature will undergo 4 months of testing before being rolled out to more subway stops across Seoul. Possibly on smartphones as well. Meanwhile, South Korea's electronics giant Samsung says it will roll out its next series of Galaxy smartphones with a built-in AI translation service in yet another example of how artificial intelligence is being folded into our daily lives. Eunice Kim, Al Jazeera, Seoul.

All right, let's bring in our panel of experts. From London, we're joined by Toju Duke. She's the founder of Diverse AI and a former program manager for responsible AI at Google. From Sydney, we're joined by Toby Walsh, a professor of AI at the University of New South Wales AI Institute, and from London, Ian Whitaker, managing director of Liberty Sky Advisors. Welcome to you all. Toju, let's start with you. As AI grows, its appeal is undeniable. It's no doubt here to stay, but is it a good thing? Is it going to benefit each and every one of us, or is it going to benefit a few and make a lot of us much poorer?

Well, that depends on how it's being deployed and being developed. If it's developed properly, responsibly, and using responsible AI principles, it will benefit everyone. At the current stage that AI is being developed, it's been moving really, really fast. The industry has this mantra of move fast and break things. And I think the mindset of apologize later as well. And with that sort of mindset, a lot of people will be left behind. If we change the wheel of motion, right now and change the pace that we're developing AI technologies, especially generative AI technologies and AI technologies that seem to be surpassing human intelligence, with a big question mark there. And if that whole mindset is changed and we, we kind of like develop these systems a bit slower and bring in everybody on the table in part of— as part of the decision-making processes, the design, the development, and testing of the impact on on all the users across the world, then it will benefit everyone in society.

Toby, what's, what's your view on this about whether AI is a good thing? Is it going to be good for humanity? And we were all sort of getting familiar with it now. We're encountering things like ChatGPT, uh, it's, it's in some of our cars. How much more will it influence our lives going forward?

I think in the future it's going to be hard to imagine our lives without artificial intelligence. But it's about us making the right choices now. This is not the first technology that's disrupted our lives, and it's about ensuring that the wealth isn't concentrated in the hands of a few. And I think there's increasing recognition among civil society, among politicians, amongst the public, that just as the robber barons initially captured the wealth of the Industrial Revolution, that we need to make sure that these digital roger robert barons, that also capture the wealth of this current revolution.

Ian, you're an AI skeptic, if I can put it like that, at least in terms of the amount of money that's being spent and invested in AI and its practical uses. Is AI just a flash in the pan, or do you think it does— do you agree with Toby that it has the potential to impact all of our lives and we can't— we won't be able to imagine how we ever lived without it.

Well, funnily enough, I think what you'll have is you'll have both. And what I mean by that is that you're right, I am a skeptic. My view on AI is that, quite frankly, it's very much— it will provide the average. May provide average good, may provide average bad. There's many particular roles, and I think your report highlighted that, where quite frankly AI just sort of won't be able to handle the tasks that humans do at the moment. And it's very hard to see whether that will change in the immediate future. I think that when it comes to AI and use of technology, and the way I see it is really that AI will be used really more as a cost-saving measure rather than necessarily as revenue generation. And this is why I'm skeptical about the whole idea that it will necessarily add to the value of the world economy. I think many firms will use it to reduce jobs. And you're seeing this in areas, for example, such as chatbots and customer service and so forth. Well, I think the applications in which the idea that artificial intelligence would drive new revenue growth coming through from firms, I think that's quite fanciful. And the way I would look at it, quite frankly, I'd highlight two things. One is that with, as with a lot of technologies, what tends to happen is that people underestimate changes in human behavior from actually implementing the technology. For example, email, where all that happened was that people produced more and more email. But didn't necessarily lead to an increase in productivity off the back of that, just simply because sort of people's behaviors change. I think the second thing is look at what's happening, for example, in the US military, which has been a big proponent of using AI over the past few decades, arguably as far advanced sort of as its civilian counterparts, is now actually pulling back on AI because it said it just doesn't really deliver. What it is hoped for.

And as far as markets are concerned, Ian, do they really understand the potential pitfalls of AI along with its benefits? Is what's going on at the moment with all this investment and the way that markets are excited about AI just another bubble?

Well, I think you can see that markets and investors don't really know by just the reaction that they have when there are relatively minor events. When it comes to AI demonstrations. For example, if you take when Alphabet had a hitch in its presentation on AI, over $100 billion was wiped off the market capitalization of Alphabet in 2 days. And yet that is an extreme reaction. But what I think it shows is that, quite frankly, investors and analysts at the moment don't really have any idea whatsoever of what could be the potential of AI at the moment. It's very much driven by an element of huge sort of hype, I would argue, but also as well that when it comes to the fundamental technology, there's not really a big amount of understanding within the financial markets.

Toju, what do you make of what Ian was saying there? Should the development of AI be unfettered? I mean, currently there is very little oversight over how companies are integrating AI into their interactions with us, the public. As an advocate of AI? Do you believe that it needs firm regulatory oversight in its development and uses, particularly now in these early, early days?

Well, AI has been here for a very long time, over 200, um, 20 decades ago. And, um, we've been seeing lots of AI winters and summers over the years, right? So this bubble that we're talking about is not the first time we're having this sort of bubble where, you know, AI promises so much and investors get really excited, pump so much much money into it, and then people, the same investors and, you know, organizations, kind of like, um, lose faith in AI because the capabilities and the promises were not delivered, and they take the money out of it. So right now we're back in an AI spring, on AI summer, some people say. Um, regulation has been here for a while, like, you know, for the past couple of years we have the EU AI Act, we had, um, Joe Biden's, um, executive order on AI, and a lot of countries across the world, especially in the western side of things, have been developing AI regulation. Um, so back to your question to Ian, like, you know, how much do investors really understand the pitfalls of AI? I think it's really small. I think there's a very lack, um, general lack of understanding of how AI can improve people's lives, and there's a lot of excitement around the business promises of AI. Um, so it's, um It's almost like a chicken and an egg problem, right? If we're not bringing people along and bringing communities along and thinking about, thinking beyond the business impact of AI to the benefits of AI, and that's why we're having issues where people are scared about job displacements from AI and scared about AI just, you know, taking over their lives and they have no form of control or input into it. And if we don't change the wheel again, like I said before, there's going to be a lot of harm to society and we're already seeing that now. And there's only so much regulation can do because there's lots of politics politics involved in this. AI is really boosting the economy. A lot of, you know, governments are really interested in the capabilities of AI. But at the same time, there's lots of harms that's happening to people and people from marginalized societies and young people as well, vulnerable people that people are not really getting to hear about. And some of it makes it into the media. For example, there was a teenager from Florida a couple of weeks ago who took his own life because he was having a very close interaction with an AI chatbot and an AI character. And that's not the first story we've heard about this. And regulation still exists right now, but it's not able to stop these horrible things from happening to people.

Toby, do we need to regulate AI? I mean, what are the dangers? Toju Duke highlighted some of them there. What are the challenges for lawmakers as they deal with a technology that holds such great potential but also poses such great threats? And also, how do you regulate an industry which is growing so rapidly and changing so rapidly?

Well, 100%, we do need to regulate AI. You know, the largest companies market capitalization on the world today are tech companies. It used to be that they were banks and oil companies. Now they're tech companies. And large businesses always had to be regulated. So, yes, we do really have to regulate them. And it's interesting to see, you know, we are already starting to see the beginnings of that. The EU Act is a good example of what we're going to start to see, which is that politicians and the public are waking up to the idea that this is very large business and for the benefit of the public, it needs to be regulated. And there's immense financial pressure. It's $1 billion a day that's being invested. OpenAI, the company behind ChatGPT, fastest growing company ever in the history of capitalism. They went from no income to one year later, $1 billion annualized revenue every year. We have never seen such spectacular returns. But equally, there are significant risks. The risks are very obvious. The risks to our democracy are that these companies cross national boundaries. They have the wealth of small nations. They could disrupt our very electoral process. So, yes, we definitely do have to regulate. And it's, it's easy to do so because, you know, they only get to act in our economies according to the privileges that we give them.

But Toby, will regulation stifle innovation?

Well, that's what we thought would that if we were careful, it wouldn't happen. If, you know, we didn't— you know, we've, for the last 20 years, we thought if we regulated, it would stifle innovation. That was probably true 20 years ago. It's not true today. Indeed, if we don't regulate, we will stifle innovation in the future, that the public will turn against them, that we will have so many bad outcomes that it will stifle innovation. So indeed, I think many people— indeed, you know, I know many of the people who work in Google, work in Microsoft, work in OpenAI, they would welcome some regulation. At the moment, it's a race to the bottom. They would actually say, you know, we don't do these things because we choose to do them. We do them because our competitors are going to do them. Someone needs to set a floor in these behaviors, and they would welcome some more regulation.

Ian, where do you stand on regulatory oversight? Have governments been too slow in at least taking a view?

I don't think that's necessarily the case. I mean, you've got— you do have a lot of change that's coming through and things do seem to change almost on a weekly or monthly basis. I think a lot of the issue will be, quite frankly, governments and regulators not having a proper handle on what the implications are of this technology, not really understanding it. And this is something that you've seen when it comes to the tech sector as a whole over the past 15, 20 years with the likes of the rise of Google and Facebook and so forth. And so I think there is definitely a knowledge issue there that you've got when it comes to regulation. I suspect actually what will happen is that in some ways with this technology— and it goes back to what you were saying in the report— the markets in a way will regulate themselves. And what do I mean by that? Well, I think it will become pretty clear to markets over the next few years and more, I'd say, to businesses what AI technologies work and which, quite frankly, are a waste of money. And I think there are some areas that I'd say has huge potential impact, for example, scientific publishing and the ability to go through masses of data in a very short amount of time in order to come up with new medicines and new applications. And I think most people would see that as a positive to society. I think the ways that many companies are thinking about AI at the moment and promising sort of the promises they're making in terms of how AI can improve things, I think are vastly overhyped. And I think what will happen with those companies is there'll be an initial spurt and investment will come through. And when they realize it doesn't work, then what will happen is they will back off themselves. And what that will lead to is a situation, I think, As I say, over a number of years, what will happen is AI will get channeled into certain areas where it does have a major impact. But in other areas where, quite frankly, its impact was overhyped, then I think the funding will stop. And to go back to the point I made before, we've had two back in the early '70s. That was the AI winter caused by the oil crisis. In the late '80s, that was caused by the Japanese economy faltering. But when money is pulled from AI because it demands so much money in order to push through all the changes, then what will happen is AI by itself sort of progress in certain areas will stop.

Toju Duke Duke, Toby was talking about one company in particular, the company behind ChatGPT, almost dominating the field at the moment. What about in terms of national interests here? Who's going to win the race to dominate in AI? Is it going to be the US? Is it going to be Silicon Valley? Is it going to be China? Is it going to be someone else?

I think, I think the answer to that is quite obvious. I mean, Silicon Valley has been leading in this race. Most of the AI companies are all based in Silicon Valley. With the AI Act coming out, there was a bit of a problem with it, you know, and it almost got deprecated because, you know, the people in Europe were complaining that if we keep on coming up with these regulations, Europe doesn't stand a chance against the US because the US is not coming up with such hefty regulations to stifle innovation. Um, the EU AI Act is still there. It was changed a little bit, and I think it's a bit more friendly. Um, but I mean, part of the problem we have is, is just the concentration of AI companies coming from the US. China is also doing a lot of things. We don't really get to hear so much about what's happening in China, but we know there's a lot around computer vision and facial recognition and, um, You know, the people in China, from what we hear, do not really have a choice with the AI systems that are being shoved into their faces, literally. In Europe, we have very few companies that are working on AI, and part of it is really due to the investments, right? Especially when we're thinking about generative AI. I think it's important to see the difference between the traditional AI, AI as we've known it in the past, which is still working today, like predictive machine learning, you know, using it to forecast data analysis, AI has been here and people have been using AI. Businesses have been using AI in this format. Then generative AI came along from OpenAI, and then the AI arms race started and all of the other tech companies had to, you know, show, show some value to their investors. And there was competitive pressure. And generative AI is coming up with so much promises that they're unable to deliver. And that's where the problem, problem really lies, is there's so much expectations around generative AI. I believe generative AI at this stage should still be in the experimental form. It should still be in the research labs. It shouldn't be put into enterprises just yet. People can still run pilots with them and test them out, but do not put so much investment into them considering the amount of investment they require to build one large language model, for instance, which could cost over $4 billion or $4 million.

Okay, I'll be back with you in just a moment, but here's something else to think about. All this emphasis on AI may have an impact on the environment. Electronic waste, or e-waste, is projected to increase 1,000-fold by 2030, largely due to the demands of AI. That could be as much as 5 million tonnes of material that's difficult to dispose of. E-waste is already at record levels, much of it toxic, and only around 12% of it is currently recycled. Toby, that's an issue that many people won't have thought about really with, with, with AI. They focus on the positives. I suppose this comes down again to regulation, doesn't it? The impact that AI is going to have on the environment. Should there be individual state-level laws that are put into place, or should there be a global agreement on what happens not only with the development of AI but also its environmental impact?

I'd be worried about a global agreement. It's very hard for countries to agree. But yeah, we should be 100% worried about, about all these externalities, the environmental costs, the energy costs, the costs, you know, there are also the costs of collecting the data. Yes, these are things that we should be worried about. These externalities the market are not properly pricing, these are things that we really have to care about because they're going to have long-term impacts upon the benefit, the net benefits that we get from AI or not.

Ian, how worried about are you about about AI? I mean, we're talking a lot about the positive impact that AI is going to have on all of us. How concerned are you particularly about this issue of regulation and whether governments are moving fast enough to get a handle on its disruptive potential as well as its positive ones?

Well, I mean, from an AI perspective, as I said, I'm quite skeptical. I mean, I'm not hugely worried about generative AI as a technology. I think there are definitely implications from it and it will have implications on society. I mean, to go back to the last point in terms of the environment, what I would say is you could say, as Toby highlighted the negatives, you could also say the positives. What's happening with AI is actually turbocharging innovation in areas like nuclear energy and forcing tech companies to get more involved and to put in investment. And I think actually that's a positive for society given the issues that we have with energy infrastructure. My feeling is, is that where the risk— the risks actually are at the moment are overreacting. There's always a tendency and we've seen this before, sort of, of we have to do something now. I think with AI, I don't think that's necessarily the case. What we can do is that we do have time to see how things develop, see what are the things where there are sort of potentially the most issues.

Okay.

And the things where, quite frankly, fears have been overhyped. And what I'm concerned about is that if we sort of have this idea that we've got to rush to regulation now and at this point, what will happen is we'll lock in regulation. It will tend to be long lasting. And then what will happen is a few years down the line, people realize all the flaws in that regulation. But but it was never seen at the time.

I'm sorry, we're out of time. Many thanks indeed. It was a fascinating discussion. Really enjoyed that. Toju, Toby, Ian, many thanks indeed for being with us on Counting the Cost. And that's our show for this week. If you'd like to join the discussion on AI, I'm @AFinnegan on X. Try to remember to use the hashtag #AJCTC, or you can drop us a line. Countingthecost@aljazeera.net is our email address. As always, there's plenty Plenty more for you online at aljazeera.com/ctc. That takes you straight to our page, and there you'll find individual reports, links, and entire editions for you to catch up on. But that's it for this edition of Counting the Cost. I'm Adrian Finnegan from the team here in Doha. Thanks for being with us. The news on Al Jazeera is next.