The AI Republic | Mark Esposito | Talks at Google

Good afternoon. So the way we designed this opportunity for this afternoon together is to have me first tell you a little bit about the, the Arab Republic, which is like the umbrella under which we're going to talk today. But this is pretty much a Q&A opportunity for us to talk and to really exchange on an important topic, which is not only artificial intelligence, but the implication that these technologies and this architecture technology are going to have for us in the next few years. So I'd love for you to ask me questions and making sure this becomes a dialogue between us. I'll still give you a little bit of the story behind The AI Republic. So with my co-authors, Terence See, who was one of the two co-authors of the book, back in 2015, we started the journey about understanding megatrends. and we wanted to understand what is currently happening because we were starting to notice a lot of anxiety and angst in the world. We tend to teach business leaders, and so we tend to see that a lot of their pressure was coming from not really knowing what's going to happen in the years to come. So from there, we started a journey that brought to build a network, a framework called DRIVE, which was a framework that was supposed to really capture the idea about what are these trajectories that are rapidly changing and they're creating a lot of change in our society as well. So DRI stands for demographics, resource scarcity, inequalities, volatility, and enterprise dynamics. And after these 5 trends that we studied and making sure we understood how we can help organizations and governments to navigate these exponential changes, we started discovering that technology was becoming way more than what technology has always been before. Before, technology was a means to an end. It was facilitating the efficiency in any given process. And today technology was becoming an architecture that by itself did not need to have people to operate, didn't need to have space to operate. And it was to some extent globally distributed, 'cause we see more and more country trying to use technology solution to address some of the challenges we have. But this technology was not wholly creating impasse like the displacement of labor or the fact that some of the traditional institutional factor like ledgers were now being shifted over to smart contracts or things that resemble blockchains. But there was also a fundamental change in what we think this technology will turn society to be. And we are living in, and as you feel it as one of the leading companies in the world for technology, this is really the very, the very midst of a transformational period that will change forever everything that defined the 20th century institutional frameworks. So we got intrigued and we wanted to go deeper with that. At the same time, as Sugul has introduced, we co-founded a company called Nexus Frontier Tech. Like every good story, which is good to tell, we started with nothing. And now, as she mentioned, we have 130 employees, mainly based out of Vietnam. So we have an academy called Rubik AI. We mainly train mainly women to become coders, and then they stay 3 years in the academy. They work for us for about 1 year. and after 1 year, they get headhunted for x15 of the salary we can pay. That's normal. It's part of the problem we have in the industry that most of the talent is going in the hands of the very few companies that can afford to have talent. But we know this from upfront, so we understand the turnover will be very high. So Nexus became for us a bit of a social incubator, but is equally a quite successful company. We are a real AI company, which means that we have real AI scientists. Our entire scientific background comes from Japan. The Japanese have been ahead of the game since, you know, we even started to understand what is AI. So they are 20 years ahead in the conceptual understanding of AI. They don't have the financial capacity to turn into organization like this or any sort of organization that is AI-specific. So for them, it's more of a cultural paradigm than anything else. So they are perfect in training, mainly the Vietnamese coders, to understand AI from the perspective of conceptual models. Then we have one pocket in Bangkok. We're now opening one in, in Kuala Lumpur with a small pocket in Singapore, but our office is in London. So we started as a small, small company. Now we're shifting into a bit larger company. We are like every company that wants to survive, trying to expand our financing. And we are on the route to IPO 2022, 2023, most likely NASDAQ. And so we've been working with KPMG to get underwriting so that we are ready for this. We restructured the boards. So it's one of those stories that follows any form of technology company story that eventually, if we're not acquired, will go public. Now, what was the learning from this? You know, we're not coming from technology. As Segula introduced, I come from economics. My co-authors came from finance, and Danny is a technologist, so he came from technology, but he's not a scientist, so to say. So we started to say, well, why do we— while we understand technology becoming more predominant, we equally understand that artificial intelligence as an architecture is becoming the most interesting story to tell because it's the one that is integrating multiple technologies. Is within the spectrum of what we call the Fourth Industrial Revolution. And this was something that was reinforced by my experience with the World Economic Forum, which coined the term in the first place. We started to see that this integration of technology, they are interoperable and compatible because they are converging. And as they're converging, they're creating this intelligent neural system that was mainly replacing some of the most repetitive tasks we used to have before. So we start to notice that this technology is becoming predominant, but it also starts with a number of different strings attached. Number one, this technology is not distributed. The conversion between technology development and general-purpose technologies is quite low. It doesn't become shared across different code of mass customizations. We notice that the technology companies that are mushrooming around the world They tend to have a rather short lifespan and they get acquired by the traditional players. There is almost like a systematic conversion from startup to acquisitions. And interesting, they're not acquired to preserve the company ideas. They mainly acquiring the asset, but most likely the people. We start having overdependency on very few companies. And where is the rest going? If it's not enough from a geopolitical perspective, With the exception of the US and China, nobody really is competing for AI. The US is in the private sector. It does not necessarily scale down or cascade down to the private sector. Many American cities are still somehow attached to the 20th century design. So we haven't seen the conversion between American technology penetrated into the livelihood of people. But in China instead, we have seen that government government activity is shifting AI to be almost like a national asset, ensuring that from commerce to IoT sensors, security, surveillance, and of course artificial intelligence, we have way more, way more presence in China in a number of different ways of AI than what we noticed before. So we started to write about this. So we took about a year and a half. We wrote with Danny and Terence, The AI Republic, We have one major purpose. Number one is to tell everybody AI is not intelligent. It's just a computational capacity that has the ability to replace a lot of our jobs because our jobs were designed when none of this technology was there. Number two is much more about intelligent automation rather than just intelligence. The term was coined in 1956 for the first time. It was a way to define some form of cognitive skills, but we are far from thinking that it can think. Of course, it does good enough to make us look relatively limited in the amount of possibility we have to preserve our socioeconomic structure the way they are. And many of our jobs are somehow related to some degree of repetition, not necessarily here at Google, but of course, outside of this environment, you notice many jobs are still designed with 1950s or like job models. And interesting, if I'm asking the question, how many jobs existed in 1950? And I have an answer that says X. And how many of these jobs still exist today? The answer will be that about 90% of the jobs that existed in 1950, they still exist today. So naturally they get exposed over time to be replaced. So the social implication is much more profound than the economic one somehow. The economic one is way more important than just the technology one. And governments have been quite absent in understanding technology as an opportunity to readjust social gaps. So, we would like to think the government maybe can start stepping up because with the exception of very few governments, most of them very small and very entrepreneurial, most governments still struggle with the idea of technology as if technology was still just the IT department. Which as you guys know is not. And that started to be our journey. So we wanted to think, can we think of technology becoming much more diffuse and distributed? We actually had a working title, which was, There's No Intelligence in AI, but our publisher said, if you guys want to sell this stuff, you need to make it more appealing. So we started to say, well, what could be a sense of distribution of technology? So we talked about the AI Republic. The idea that technology is now shared and diffused and available to everybody. Because going back to one of the major concerns that I have and that I'd like to share with you, which will also finish my preliminary conversation before your questions, is that if we're not distributing this technology across, very few countries will have control across everything we do. The problem with that is that it creates not only ethical issues, But then it changes forever the nature of geopolitics, equally changed the nature of democracies, and it changed the nature of behavior because as much as we do research about behavior, we notice that no matter how special we think we are, we're quite predictable. And our behavior tends to be somehow summarized by pattern, default mechanism, decoys, context, social norming, peer pressure. So as much as I I still like to believe the story from, from a perspective of societal value, we tend to be quite predictable. So what will happen when companies will be able to know everything about my life better than myself? And how does this change society? So this was our original conversation. Now, the paradox is that we have an AI company, so we have all interest to be advocates and evangelists for AI. But we are also not scientists from AI. We're mainly socioeconomic scientists, right? We are, We are mainly school professors, business school professors. So how do we make sure that we are navigating this conversation where we have an implication for the economy, for governments, and all of that? And that's the nature of the book. As Segoul has introduced, the book went very well. We were lucky that we were able to get to Amazon bestseller status within the first few hours of the book launch. It gave us a lot of opportunity to engage and meet people. So beside the teaching and writing, most of my time these days doing book events or book conversation like the one we have right now. But this was particularly important to me, number one, because my constituency today is an important one. You work with technology every day and you are defined by technology more than other companies. You probably have dilemmas as you see as people, the evolution of this technology. You must find yourself that as a global company, compliance operating in multiple contexts is becoming more and more challenging. You know the government can accelerate but also shut down things right away. And we have lack of governance on technology. We currently don't really have a blueprint. Should this technology governance be orchestrated at the international level, at the multilateral organization define money? Should we think that we should operate like the banking system? Should we think about regional cooperation like the EU was the original plan? There was a model of super parties that was interested about regulating the interest of the region? Should we think about inter-regional cooperation that could happen, for example, in an area like Asia where many countries might need to orchestrate their resources also from technologies? Or should we just let the market decide? Should we— where is the, the boundary between market validation and innovation graveyard? How much can the government decide this is something I want to control because it's in my best interest, and how much will that eventually lead companies to leave and go elsewhere. So these are some of the dilemmas that I guess you face, you ask yourself, I ask myself. And today is just an opportunity before— go figure, we get all quarantined for some reason— to have this conversation, talk, and get maybe the opportunity to meet each other in this comfortable setting as we are. So that's my little intro, and I'd love to hear more about questions or comments or things you'd like to share. The more we make this a conversation, I think the better it will look like, right? Especially on video. So let's start. I would say, like, any question? Yes.

So you mentioned about countries and governments taking the lead on AI and the examples you gave Germany versus China are very different. Sorry, US versus China might be quite different. From especially from, from the point that where they're approaching AI, uh, participation versus control, uh, innovation versus, uh, you know, efficiency and so on. Um, from your work in, in APAC, do you see that there is a chance that AI can be democratized and can follow an open source kind of a I don't want to call it a revolution, but a movement and sort of counter or at least match a government force on the field?

Thank you for the question. What's your first name?

My name is Asla.

Asla, thank you for this, Asla. So I think first of all I have to tell you what prevents AI to become applicable Imagine that we could. Number one is that we still have to digitalize a lot of stuff. There are many organizations and governments that still operate like 1970s, 1980s. They have not modernized the processes. They still are stuck into control and command. They still do not trust. They need 5 steps to approve a process because you need 4 people before you to trust that the 5th person will say, yes, let's do this. So I think we, especially from the entrepreneurial side of my job, We notice that many companies don't have the ability to digitalize fast enough. They come up and say, we'd like to have AI, sure, but their entire value chain is still quite analog. So how do we do that? I think the first question is, how do we recognize opportunity for technology to really correct the inefficiency in what companies and entities do and help them eventually discover a higher possibility for them to position themselves in the market without feeling that they are left behind. Once you have that step, the next step we find is the question on what do I need in terms of the type of needs I have? Most of the time, they don't need AI. They need some form of software, and that doesn't have to be even intelligent. You have to be smart enough to make sure that it is autonomous. but it's not AI. Some of them start having interesting questions that reply predictive analytics or predictive modeling, then machine learning kicks in. But how many of them really require AI? A very small part. So I would say we might get to an AI model when we start digitalizing more and more, when we're having government funding helping modernizing the infrastructure that we have in business. Then I think we can start asking, can this become open source? Will this become open source? Well, so far, the country that control these technologies, they don't have many incentive to eventually make it available so far because there's still a lot of value in proprietary technology like the one we're talking about. Furthermore, shifting from traditional asset to data made it quite interesting that if you have control on the dataset, you pretty much control a lot. So I think we have to start thinking what is the economic incentive for any entity to distribute these technologies? The only answer I have is to be driven in collaboration with governments, as government interest would be to make sure this technology is becoming distributed because government need to figure out how will we modernize an economic model that is now becoming obsolete. Technology like AI or anything related to AI has the potential to create jobs we don't have. But we have to start suspending some of our usual suspects. Number 1, technology versus human, who is more efficient? That's a quite dumb question to ask. So you cannot run it as an equation. You have to run it as some form of combination between human job and technologies. So in more physical job, we call it cobotics. You guys know this. In less physical job, we call it symbio-intelligence. It's the ability to integrate technology in what we currently do. And by doing this, you're generating a spectrum of skills that currently we don't have. I think that would be a lot of value. We can create millions of jobs that we currently don't have without fearing that all we're trying to do is to bring manufacturing back to a country, no mention of country whatsoever, or to make a country, you know, again predominant again, because that model of nationalism will never work in this day. So I think it's really about, first of all, multiple steps required. Before we're going to get to that point. Do I see this happening? I see this in small countries that have not a lot of historical heritage, this desire to modernize in the future. But in places like the US, I'm not talking about the technology companies in the Valley, but in, say, let's say a country and in the EU, I still see so much attachment to the past. There's more interest in reviving the past than looking forward to the future, unless you're Ireland. Therefore you've got nothing to look backwards. You can only look forward, right? But go to the UK, and for example, the UK is a perfect example of being trapped in this mirage of the past. You know, what used to be an empire now is going to be an island isolated from the rest of the world. But that's my political stunt. So, but thank you for your question.

Hi, my name is Ron. Um, two-part question. First of all, how different is the AI revolution from the industrial revolution or the invention of electricity in terms of an impact? And the other part of the question is, if it's just about replacing mundane, let's say, or jobs that can be automated, why not just leave it as it is? What's the risk of maintaining the status quo between US and China?

Thank you. Beautiful question. And beautiful because I think they help our conversation, right, to, to go to the next level. I'll start from the second one. The issue with keeping things as they are is that a number of people are difficult— it's difficult for them to get trained to do anything else. Why? For too long they've been doing the same job and their skill gaps become wider and wider up to a point in which you can't train anybody anymore. Take, for example, a petrochemical engineer. With a degree, if you're replacing that capacity with an automation, there's not much that this expert can do outside of that context because he is so deep in his knowledge or her knowledge that transferring over is a significant cost. So one of the problems we're going to have is that a large part of the population will become unemployable. If that happens, it's not the issue with unemployability, it's the fact that when population loses economic incentives, it's easy to radicalize public opinion. And when you radicalize, in history, monsters show up. We saw this when we saw the less integrated part of the UK decided to vote to leave. They have less at stake. The most forgotten part of the United States, the Rust Belt, the one that was less impacted or least impacted by the prosperity of the globalization decided to vote against the establishment. In parts of the world, like in Brazil, the constant discontent that people have allowed nationalism populists to rise. Again, just to be very candid, the problem that I have with nationalists is not nationalism per se, because it's perfectly fine to look after the interests of your country. It's when nationalism is understood as a contrast with globalization. The best nationalists are the ones that actually are globalists. Because the only way you have to defend your own interests is to play in the global world, right? So I think that's my problem is that those kind of polarizations, not only they polarize public opinion, but they give space to people to use rhetoric to merely enchant people. You know, when I'm going to say, for example, and I don't want to have a reference to any country in particular, but if I'm going to tell you that the job that I have lost to technology will come back, I am lying because no job lost to automation will ever come back. But if I have lost my job to automation, I read that maybe my factory moved over to Mexico, to China, or somebody else, I kind of believe that maybe something can be reversed back. So I find it difficult that if we're not addressing the current situation, we might have even more discontent. And discontent is an engine for a lot of issues, so to say. To the first question, which is a beautiful question, In every other revolution we had, the displacement of labor happened, but we were shifting people from one simple job to a more sophisticated one. So interesting, if you look historically in the previous three, every time that we move from one level to the other, more jobs were being created. So if I was working in a farm and now I'm having, you know, maybe a tractor that does the job, many of these people leave the farming and they go to the cities where they're now becoming mechanics, right? So that evolution has improved their position in life. They shifted into a different level of economic mobility, and they start planning their future by saying, I want to get married, I'm going to have kids, I will save for my kids, and my kids eventually will go to university and all of that. So the previous three models were always, I displace, but I'm improving the position. In fact, if you look, measuring the number of jobs created historically during those three, every time we displace, more jobs were being created. I'm going to give you one simple example. When the ATM was introduced in banking in the '60s, people working in the bank were freaking out. They were saying, if now you have an ATM that's doing my job, will I lose my job? Well, no, it's because of the ATM, the people in the bank could do things more significant than maybe just counting money. So ATM has allowed the banking industry to prosper even more. So imagine this applied to every single industry before. So it was actually a blessing. Now, that's actually the catch. By displacing the job and by creating efficiency, I do not necessarily create new jobs because you can have a dark factory that runs— it's a factory running without people, mainly robotics. That is called dark because you don't need lights when you don't have people. That is as efficient, if not more than a normal factory, 'cause now you're working 24 hours a day, where the revenue of the firm tends to be quite high, but that specific revenue is not redistributed back into anything other than taxes that the company pays. There's no jobs in the same proportion that you actually are displacing. So that's why the Fourth Industrial Revolution is a blessing if you embrace it because you modernize the way you think. It's a curse if you don't embrace it because the level of gap that you're gonna have in the future will become unsurmountable. You might have country, company, or entity that will never be able to catch up. So I think that's the main difference. Reason why we have to be vigilant now is because by itself, automation make us quite redundant in most of the job we have, unless we redesign in education, unless we start thinking that the way human labor should be measured should be with KPI that reflecting nonlinear distribution. Unless you're trying to think that we are going to be used in non-repetitive tasks across the system. But I know a lot of jobs, not necessarily at Google, but when I'm talking this to a class of executives and I ask the question, how much of your job is repetitive? And all it takes is that you write in emails, PowerPoint, and spreadsheets. If they say 30% to 40%, I always say the economics is much more in favor of automation than in keeping your job. Because bear in mind that in a company, on average, labor cost is about 33% of the entire cost of the company. So somebody, when they say save money, the easiest way to do is to replace it with automation. Now, people will say at first it's expensive, but you can easily demonstrate it after 12 months. The operating expenditure, the OPEX, is now paying off, and now you have profits. So I think that's the danger we have. But thanks for your questions.

My name is Walid. I want to go back to your point around the danger of AI and technology being controlled by a few companies and a few countries and contrast that versus the developing countries that are dealing with the basics of having electricity, feeding people, and so forth. So the gap is huge and the gap is only accelerating. So how do you see this playing out? Because as you mentioned, the developed countries and so forth, their behavior is not very benevolent. And so how do you see this playing out? Do third world countries and developing countries have an insurmountable task? Are they really behind now, or how will this kind of— how will they benefit from the tech and AI?

Thanks for this, really, for the question. So I find that we have examples of countries that 25 years ago were fighting for their life, like Rwanda, and now it's one of the most internet-penetrated countries in the world. They start having blockchains in part of the government functions. And if somebody would ask us 20 years ago, can this country ever become a major prominent country in Africa, it would be difficult for us to say. Right. We've seen some specific areas, areas like Morocco becoming a leader in renewable energy and some specific kind of manufacturing because they understood the energy and energy storage is key. So a lot of the energy that used to come, for example, from France now come, for example, from Morocco. In countries like Spain, they have been struggling with energy efficiency for quite some time, or Tunisia following similar trends. So I see examples of countries that they were able to leapfrog because they have nothing to hold them back. And I think this is like my answer to your question. If you have a lot of inherited structure from the past, it's difficult to change because you start having a lot of resistance from the fact that you have an inherited structure and 5 generations of people at work. You know, people that are at the end of their career, they don't have the same thirst and appetite for technology as somebody who just started. But because of life expectancy and longevity, we now have literally 5 generations of work plus robotics and AI. It's kind of a complex environment to have. But then to play up a bit on your question, many of these governments, they get loans, they get funding, they want to attract investors, but they still attract investors that speak a 20th century rationale. So they don't bring technology companies over to help them build an ecosystem. They mainly build being a company that brings real estate or oil and gas. As much as they are profitable businesses, but they do not really reflect an image of the future. So I think it's how do we help them and educate them to think that the investment in the country to start building an infrastructure that become a threshold for the digitalization to come through, for the automation to come through. I think this is the way forward. I don't see this happening immediately because I find, and to be honest with my answer, in some parts of the world, the political class is very detached from the reality of the country, or they're so old in comparison to the average age. One of the first things I say when I'm teaching more geopolitical content is in the African continent, we have probably one of the best representations of average age 20 years. The politician running the country might be 99, 94, 89. So we truly have a gap in the priority that you might have in making decisions. I think this is somehow one of the challenges I see. So to finish on that, if they don't catch up, they will become what Harari, the guy who wrote Homo sapiens, called data colonies. No jurisdiction anymore. Digital sovereignty means that sovereignty will be in the hands of very few companies. I let you imagine how much power you have if you control every single aspect of an economy because it's now a digital data set. So I think this is the risk that we're running, and I'm not sure this is something that will happen because the estimate in things like this always demonstrate that The difference between the estimate, the expectation, and the forecast and the reality might be very different, but it's a risk that we're running. So how do we encourage the right level of investors coming to those countries? How do we help the political leaders to understand that technology is no longer just efficiency in the production, but is a way of defining competitive and comparative advantage? Multiple conversations to have, but thanks for your question. Yes.

Hi Mark, my name is Ryan. My question is around— you talked about the digitization of data as a step before getting to AI models. So as we digitize the data, there's a lot of bias in that data because we're creating it as humans. So are we going to get to a place where we actually teach AI models ethics and values? And if yes, whose ethics and whose values?

Good question, and beautiful question. So first, first of all, yes, the number of data that is mainly being somehow discovered is more and more visible. I think we are discovering today two things. There is the explicit biases, which are the mistakes that we discover, like in the case of the facial recognition, just because you have a different dark skin connotation, the algorithm does not recognize you. That's easy to fix. But there's a lot of implicit biases that really are about judgment, and I think you are reflecting exactly the problem. I will have a problem with the cultural predominance of a given group of people. When Sugu was introduced in my company, one of the reasons why we pride ourselves on having 80% women as coders is because the nature of the bias we're going to have, first of all, will be very different, but also much more mitigated from the bias of the traditional engineering culture. So I think one way to eventually mitigate, because we can never solve biases, biases are part of human behavior, is to mitigate by having a number across cultural teams, they're able to defeat the predominance of the mental model. If I have 5 engineers in the team, I will keep on having a bias that is coming from their mental model. If I start having a diversity in the talent I'm actually employing, eventually I decrease less dependency on one mental model. We saw this, for example, with Singularity University, which is powered by NASA. This guy asking people from all venues of life to to come and address major challenges, and their solutions are extremely robust because they're not coming from a specific cultural set. That said, I think there are two possibilities. One is changing the hiring, making sure we're no longer hiring people that look like us. And second, diversify much more, making sure that we can train people to think about technology as a possible career, especially people that now are not part or considering that. The second part is that how do we build capacity to start discovering biases and correct it? So I have a friend who was— she's a scientist. She did her postdoc at MIT. She's very prominent. So she had to decide what she wants to do as she has all sorts of options. She can teach, she can go into coding, she can consult. So she's consulting with Boston Consulting Group. On discovering biases in algorithms and helping companies address that. Now, when I asked her, how much competition do you have? She said, I am the only one. So how do we start building capacity for almost like, you know, we used to have Ghostbusters. How do you have bias busters? People that are able to use coding as a way to discover biases, engage in a debate. Final point I would like to share with you. Many biases coming from the fact that the The coders are not integrated, which is part of your ethical question. They're not integrated into the core strategy of the company. You still go to a CEO today and ask them, do you know what your IT does? I don't know, they are somewhere and the door is locked so they can't come out. And ask the IT guys, do you know what's the core business of your business, of your company? Many of them do not really know. And in some cases, because we wanted to horizontally integrate, We outsource this to somewhere in the world. So we need to link again technology with strategy so that the technology guys know what they're planning and programming. The C-suite understands that technology is much more than technology, it's a strategic positioning. That will decrease a lot mistake of interpretation. Final example, because your question is profound somehow, it can go into multiple levels of implication. One of the problems we have in biases is in algorithms used for education in school systems where we're using algorithms to determine performance of children. But many of these kids, if they come in from disadvantages or disadvantaged family backgrounds, they will never perform well. So what do we do? By using algorithms that were designed as standards, we keep marginalizing those people who need this the most. But what happens if you're a rich kid? You go to a private school. Private schools don't use standardized testing, they use people. The same happens for if you're asking for a loan to a bank and you're in the US and you're asking for this federal score, the credit score. People that perform poorly, the algorithm will decide that you don't qualify for a loan, therefore you don't have access to an opportunity. There is a bias in what determined to be viable. But if you're very rich, you go to private banking where you have a private banker. So you have a person that will manage your portfolio. So rich kids are coached by people. Rich people are advised in their wealth by people, but poor people are measured by algorithms. That's a major problem we have in the ethics of AI. I think that we have to address, plus profiling, which is a different conversation. Which goes into, am I using historical data to forecast performance? If I do, does it mean that if I'm using historical data from a neighborhood that has historically been exposed to crime, does it mean that I will always forecast the crime will happen in the same region? So am I using technology to discriminate even more? I think that's where we have to be extremely careful. But thanks for your questions.

Just before we continue, a couple more questions and then maybe we'll do the closing as well.

Sure.

Great. Hello, Ali.

Hey.

Hey, Ali.

If I may, I'd like to take a step back to look at things in a bit more macro way.

Sure.

And if we look at technology overall throughout history and how it's contributed to us as a species, particularly since we landed on the moon and that accelerated phase we've had since then, Today we are the probably healthiest we've been ever. We are, we lead the simplest lives possible. We have the most potential possible. But that has been followed or simultaneously come with an inversely inverse correlation with happiness, loneliness as an epidemic, increased anxiety. And it seems that we tend to, at least from my perspective, whenever we have a milestone technology that arises, we tend to address it the same way. So we start regulation, we start talking geopolitics, government centralization, using it as a weapon potentially, etc.

Right.

And then even in the discussion today, case in point, what we're saying today, we always discuss it, end up discussing the economic implications of it while the social implications of it take a backseat. What do you think those social implications of AI are, knowing that all the previous milestone technologies has led toward more individualization or individualism in society? And then on that, as a follow-up question, then do we need to, or are you for or against teaching AI emotions and recognizing emotion? And then what would that have an implication on us socially as human beings?

Right, well, the social implications are inevitable now more than before, mainly because the system is clearly showing signs of fragmentation. And I'm not talking about AI, I'm talking about our socioeconomic structure. They're fragmented. If the 20th century was the age of convergence, 21st century clearly showing that the traditional political and economic actors are failing. We're having coronavirus and the Fed cut interest rates. What does it mean? More debt into the economy. Who pays the debt? Well, people that don't have capacity. Who benefits from this? Whoever already has money. So we keep on having the same rhetoric. So I think, Ali, the question is, will the system continue to run as it is? And I don't think so. We are really getting into a model in which perpetual growth that was somehow envisioned by GDP and the industrial model is coming to an end. It can't work forever. Especially with 7.5 billion people, out of which 5.5 are actively into the economy. So I think we are gonna see the end of the system as it was devised and designed. Therefore, I think the implication will be, wow. By the way, that system defined the purpose of labor. In many parts of the world, the constitution is based on the foundation of labor. But what does it mean to work in a situation where a lot of our jobs are redundant? It changed the relationship with the planet because we now know that the climate imperative is not just a buzzword, but we see this more and more. I was reminded in a class I was teaching these days that if we are really showing signs of inconsistency with coronaviruses, imagine what will happen when the system is melting down, when we're having climate refugees, when we start having outbreaks of viruses because biodiversity that are going to mainly collapse. So I think we're gonna be facing the limitation of our model as it is. Technology has the potential to redesign what is the concept of value in a society where no longer being measured by efficiency, because your model of your conversation on individualism, if I'm gonna be measured by my KPI, my KPI are all economic conversion or something that is defined performance, clearly the relationship with that and life is conflictual. My life cannot be measured on KPI. We are the least organized, the most chaotic, the least linear entities on this planet. Life and growing up doesn't follow in a straight line. Relationships don't work always according to the manual. And if you have kids, try to read the manual how to raise kids. It never works, right? They constantly misbehave. So I think we are facing the reality that technology has accelerated almost that reality moment, moment of truth, that the economics and the obsession on growth can no longer work forever. So I always say, can technology, especially when it's so powerful, help us to navigate the institutional framework of a new emerging structure? We call it the New World Order. It's a little bit dramatic, but that's pretty much what it is. Is a system that is no longer designed on humans being part of an efficiency model shaped around 24 hours divided by 3, 8 hours a day, a productive unit, but more about what it is like to be human and what does it mean to thrive, to have economy that are mainly prospering, but without having that dependency on mathematical models. And I think technology can help. On your second question, you know, I don't know whether we should teach emotion. Now you can say emotion are algorithmic. I can teach a series of emotion anyway to a machine, and eventually I'll be able to recognize things like sadness, joy, euphoria. I think the Japanese have been doing this as a way of compensating the lack of labor or the lack of young people. So I always say, question is, what do we need technology for? My bottom line is that technology is designed to maximize human well-being. If I have to, and human well-being has to be defined by society and their determinism, what is human well-being? It might be my ability to find a life made of fulfillment. It might be my ability to be considered a father, a husband before being considered an economist or a professor, right? You have to define what that specific is. And technology can help us a lot. It can help us to make injustice visible. It can make things transparent so you're not generating this red tape culture. It can make things actually accessible. It can facilitate access to people that don't have access. So I always see technology, and by default, I'm much more of a technology optimist. But I think what we're lacking is the design. What for? Because if we're using technology as we used it before, all we're going to do, Ali, is to reinforce the system that we had before by making it even more dry and acute in the implication we're going to have socially. We should never forget that this discontent we feel in the world today is mainly socially inspired by people that suddenly have no economic model, incentive, mobility. One day they wake up and they say, what do I live for? So what do you do there? You try to upset a little bit the status quo because you got nothing to hold back to. But your question, like all the questions you guys ask, are questions that could be taken forever, right? Because they are so profound. But hopefully just It's just helping the economics of the conversation. Great.

We have one last question here. Yes.

Thank you for coming. My name is Ahmed. When you mentioned that AI is not intelligent, so I just want to understand more, like, what is intelligence in your opinion, and what would make AI intelligent? Well, I don't know whether we should pursue intelligence as an objective for technology. I think technology should be instrumental all the way to achieve something that we can't achieve When I ask this question to the AI scientists, they almost ask, they almost look at me like AI is almost like not the right word for this. It's algorithm computational capacities, the ability to do things quicker and faster with more precision. So I think we should teach, I think, algorithm to be able to answer the question that we want to be addressed. So for me, much more about How do we help our ability to select and structure the problem to be what algorithms are designed to do? So that's where I think we can improve, is that we can use technology in questions that now we're not addressing. One quick example, we're not using AI for climate change, very little. You know, we could have addressed climate change, guys, a few years back with a little bit of GDP. We would have gone way past. What was the problem is that the moment that somebody thinks that climate change is, and somebody says it's not, it becomes a political ideology, and therefore everything changes. But we could use AI to generate scenarios for responsiveness. We can generate this for outbreaks. One question we were talking with my colleague Alessandro, who is also here, the WHO doesn't understand multiplication of the virus. So what they do, they regulate according by thinking that viruses evolve in linear models. But they don't, they evolve in nonlinear model. But how do we use technology to do that? I think that's what technology should do. And that's what will give them a sense of intelligence, although I don't think the term is intelligence. But I think it will give the sense of social implications. So the technology is used for social benefit. That's where I think what it is. My definition of intelligence is our imperfection is intelligent. Our ability for us as humans to generate bonds, relationship, to create and to actually to abstract ideas. You know, the fact that we can think something does not exist is a phenomenal ability for people to see that things that don't exist can become reality. That ability to converting ideas into project, the ability of suspending a decision for something that is greater than us, the ability of surrendering to something that we recognize greater than us, then that's what defines us intelligent. For me, intelligence and being human, to some extent, they're interchangeable terms. I simply find it very difficult to use the concept of intelligence to an algorithm, not because the algorithm is not good, it's just that it's designed for something very different. But thank you for your question. So this is where I think the time is against me and I have to stop. I will continue for another 3.5 hours if I could. Eventually, I will eat the banana that is waiting for me right behind. But I mean, thank you for having me. Thank you.