Endgame with Gita Wirjawan - Aaron "Ronnie" Chatterji - All You Want to Know About ChatGPT

Episode Date: August 6, 2025

In a time of rapid technological change and geopolitical fragmentation, who benefits from artificial intelligence—and who gets left behind?Ronnie Chatterji, Chief Economist at OpenAI and former White House coordinator for the CHIPS Act, joins Gita Wirjawan for a deep conversation about the real-world consequences of AI: on jobs, infrastructure, regulation, inequality, and the fragile promise of growth across the Global South.Ronnie reflects on what it means to apply economic thinking to one of the most consequential technologies of our time.#Endgame #GitaWirjawan #OpenAIAbout the Guest:Aaron “Ronnie” Chatterji, Ph.D., is OpenAI’s first Chief Economist. He is on leave as a Research Associate at the National Bureau of Economic Research and previously taught at Harvard Business School. Earlier in his career, he worked at Goldman Sachs and was a term member of the Council on Foreign Relations. Chatterji holds a Ph.D. from UC Berkeley and a B.A. in Economics from Cornell University.About the host: Gita is an Indonesian entrepreneur and educator. He is the founding partner of Ikhlas Capital and the chairman of Ancora Group. Currently, he is teaching at Stanford as a visiting scholar with Stanford's Precourt Institute for Energy; and a fellow at the Harvard Kennedy School's Belfer Center for Science and International Affairs.------------------------ Explore and be part of our community⁠ https://endgame.id/⁠---------------Collaborations and partnerships: ⁠https://sgpp.me/contactus⁠

If you can ask Open AI to quote for you. Why would you need a product of education in STEM? Such a great question because a lot of people said, look, now that AI can write code, why would I ever major in computer science? But what I say to this is there are so many people who have their undergraduate in engineering, but they're not actually working as an engineer.

Does this mean that their undergraduate degree was not valuable? No, because your undergraduate degree is a reflection of learning how to think in a particular way. And so I feel that ironically, despite AI being able to do some of these coding tasks and it will get better and better, it still may be wise for those who are inclined to study engineering in STEM subjects, to learn the fundamentals that are going to underlie all the advances we're going to see. Just because they might not be writing as much code at work doesn't mean that it wasn't useful. And by the way, they might read better and more code with the health of AI. Hi, friends. We're honored today to be graced by Ronnie Chatterjee, who was the chief economist of OpenAI.

Aaron, Ronnie Chatterjee. Ronnie, thank you so much for gracing our show. Keitha, thank you for being here. It's a great opportunity. I'm looking forward to a wonderful discussion. Thank you. I want to ask you about how you grew up. You're a well-educated person.

I'm curious as to who had better or more influence on you, your mother or your father. My mother and father would disagree. One would say the other, but both were educators. I think it's a key thing to reflect on. And as I'm a parent now, I think about what are the things that are going to really influence my own kids. And one thing I learned for my parents is you watch what your parents do.

You don't always listen to what they say. And when I saw my parents growing up, I saw immigrants from India coming to the United States in the early 60s. My sisters were older than me, so I came along later. So by the time I came along, they were pretty assimilated to America, but they still retained a lot of their Indian identity. And they were both educators in public institutions. My father was a professor at the state university in New York where I grew up. My mother was a teacher. And I learned a lot about the value of education.

The idea that, look, the reason they came to America was to advance their education. The reason that we had a foothold in a new country was because of education. But the second piece of that, which I also took from them in the dinner table conversations we would have, is what do you use your education for? Right. You know, is it to solve complex formulas, to write papers that nobody's going to read, or is it to generate knowledge in service of society somehow? I think both my parents in different ways really influenced me to try to think in that direction.

And because they had come from somewhere else, I thought a lot about differences around the world between where they grew up in Calcutta and India, now Kolkata, versus where we were growing up in relative comfort being middle class in upstate New York. And I thought, you know, why are these differences existing around the world? And what can I do in my career, in my education, to try to bridge the gaps? And so that was, they were both a really big influence. I think, you know, they'd maybe take more credit for one or the other, but they were both big influences on me growing up. Why economics? Yeah, you know, the conversations around the dinner

table were almost always social science. And I don't know if it's just accidental. My dad really believed in economics as a theory of change. You know, he had this colleague who was an Indonesian professor and still very active, Iwan Aziz, who I know is a good friend of yours. They're a friend of ours. And there was a generation of scholars from the global south who thought about economics as a way to change the world. You know, the idea that there was so much potential in these emerging markets. And even before these economies started to take off, people saw it. Professor Aziz saw it. My dad saw it. And the question is, how do you unleash this growth? And how do you make sure that it is not so unequally distributed? Because we know that where my parents came from here in Indonesia,

there's a lot of regional disparities back then and still today. And so I think that in studying economics, I saw this as the best way to understand the world through supply and demand, through a field with some mathematical elegance, yes, but also just a lot of people and decisions in it. And at least For me, that's what resonated by economics. Wow. You just made a transition from the government. Yes. And how do you find the intersection between government and entrepreneurship?

And I'm just curious as to what the consideration was, whether you thought that was the cool thing as to make a bigger difference. I mean, you're already making a big difference in the government. Well, no, and thank you. I think I got interested in public service in the U.S. government because I felt like I could take what I was using in academia. my economics or knowledge and bring that into policy. So I might be studying or doing research on the determinants of entrepreneurial success in my academic work. But then I can go to government and figure out how do we put policies in place?

Let's say from the U.S. Small Business Administration or the Commerce Department that help businesses grow. I think this was an insight that the rules of the game really matter. You know, I talked to a lot of great founders earlier today in Indonesia. And entrepreneurs, we tend to think of founders as people who do it all on their own. But actually the environment matters a lot, right? All the invisible things that you don't see called institutions. And government plays a big role in those institutions.

And so when I came back out and came back to Duke University, where I've been a professor for a long time, I thought I got to get close to where the action is, where the entrepreneurial energy is. And it couldn't be a better place than Open AI. Well, we build these amazing products. It's going to be government. It's going to be others in the private sector and social stakeholders that really determined how AI is used. And so that's the intersection that I really feel like I'm on. And that's why I wanted to go to a place like Open AI that touches so many different parts of the ecosystem.

I want to come back to Open AI, but talk a little bit about what you did when you were in a government. You were highly involved in the Chips Act. Yes. And how do you see that as something that's going to be impactful on the Americans, but also humanity? Well, yeah. It was a privilege to be the person in charge of the implementation at the White House for President Biden. a large team of people working across government to make that program successful. My role was to make sure that all the great work they were doing was coordinated, that we were thinking carefully

about measurement and evaluation, and that we were making sure that the president's vision was instantiated into the Chips Act. The Chips Act was an industrial policy. It was a way to invest in our semiconductor industry. And I know you've had guests in the past to talk about chips and the discussions of chips around the country, around the world. For me, this idea was to take $52 billion USD and invested in research and development and the construction of semiconductor manufacturing facilities called FABS. If we could do that in the United States, then it would make sure that the U.S. had a secure supply of the most important input to almost everything in the built environment, which is semiconductors. I will say, though, you know, when I did my

PhD, the lessons I got about how to do these things were maybe don't even try, you know, industrial policy, as they call it, is not necessarily in vogue in economics. There's been a long list of failures where the government has tried to subsidize some sectors, and it hasn't been very successful. It picks winners because they're politically connected. It chooses technologies that fall out of favor or aren't market ready. And so as I approached the Chips Act, I thought, I don't need just need to get this right for America.

I need to make sure that we're making sound policy, because this kind of an impact all around the world in people looking at this example. So that's how I thought about the impact. And I hope that it's going to continue to be successful. I think it's had a positive impact so far. It was supported in a bipartisan way by both Democrats and Republicans in the U.S., which is a really good sign of something sustainability. Now on Open AI, Open AI is or was and is the preeminent, the first mover.

Do you see that as something that adds more of a moral obligation in having to do what's right. Yeah, I think so. I mean, I think what's so interesting, you mentioned why the transition or how did I think about it, I've always worked in mission-driven organizations. As a professor at a university, Duke,

that I care a lot about, there's a clear mission to educate the students, to do research that matters in the world, in the government, you know, no matter what administration you work for, whatever political party you belong to, you have a mission that makes sense to your country. And an open AI was really interesting

is that people have a very clear mission. to develop this amazing technology of gender of AI to benefit all of humanity. And when I first came in, to be honest, Keith, I wasn't sure that, you know, okay, is this a slogan? Do people actually believe it? No, no, people take this seriously. It doesn't mean that we will never do the wrong thing. It doesn't mean that we won't make mistakes. But we talk about this all the time.

And there are people in the company who they spend every waking moment thinking about how to make sure that we don't lose sight of our mission, which is very important. And for me, seeing us as a pioneer in this space, it does bring much more responsibility. And I can see this through my own lens as an economist. I want to ask the most important questions about the economy, about jobs, about the environment, and make sure I give people the best answer I know how. And if I don't do that, that to me would be a failure in the job. And so I think much more about those things, about the legacy of the work I do now, than just a simple job.

And I think that comes from a mission that really resonates with me and my colleagues at OpenAIA. If we take a look at the internet, it was a real democratizer of information. But for some of us, it didn't become a true democratizer of ideas. Ideas have gotten too polarized. And there's some economic realities that are a bit disturbing with rising inequalities, editization of the existing economic order. How do you comfort? the people out there, that AI is not going to further elitized, the already elitized economic order.

I mean, I think about this all the time, you know, and I think you've written about this in one of your papers at Stanford. It's one of the things I learned in economics was that while technology can bring great benefits, it often has increased inequality in the ways you're talking about, whether it's according to skills or geography and living through the internet, adopting it myself and seeing the impact it had on the places I grew up, on the places I would later settle. I can see that in a visible way. And so when I approach this new technology, I try to take the lessons of the economics of the internet

to understand what kind of impact it will have. There are a couple of things that made me feel optimistic about it. I'll tell you about two of them. We can talk more. One of them is the character of artificial intelligence as a technology means that we can scale human intelligence in ways that I never wouldn't thought possible. I spoke recently to a bunch of ministers

from African countries. And we talked about the biggest return on investment in Africa right now. And it's got to be agricultural extension services. And you know this well from the work here and other places, which is consulting for farmers. When you can give farmers advice about how to farm their land, what seeds to use, what fertilizer to use, you can increase their productivity, significantly improve their lives. The problem is we don't have enough people to go deliver that advice. Well, here comes AI. Our ability to take that advice and scale it to many more people. This is the kind of thing

They can actually fulfill the promise of the technology you talked about And democratize access. However, if AI is only controlled by a small number of powerful incumbent companies, if it is not adopted widely so people don't understand how to use it, it'll reinforce those inequalities

that you're talking about. And so one of the reasons I think it's cool to have companies like Open AI and others and labs that you hadn't heard of before 2015 involved is really cool. And also I think that's one of the reasons I'm here in Indonesia is I just want to make sure that this isn't a secret

that we get more people to use this technology, to understand how it works, topped it well, use the API, then create things they never thought they'd create before. To me, that's at least some of the ways we can avoid the kind of future you're talking about, but we have to be vigilant. It's a big risk and it's something I think about every day.

You've talked about this previously about the fact that a significant portion of the agriculture sector in the U.S., you know, the workforce was a significant force. I mean, part of it, but it dwindled down to 2% right. Yeah. Indonesia, for example, has more than 40 million farmers.

About of 280 million people. Yeah. How do you comfort these 40 million people by way of the empowerment of a new technological platform such as AI? Yeah. That they don't have any reason to worry about, you know, for purposes of reskilling, upskilling, or, you know, reallocating themselves to others. I would say that, you know, these. transitions, even in the United States context, from farming to a more urbanized, industrialized economy did take time. And we're still very productive in terms of agriculture as well and produce a lot

of food. So I think the agricultural sector in Indonesia is going to continue to be strong. But I would ask those farmers to think about the kinds of careers that their children wish to have and how to prepare their kids for that future. And some will say, look, I want my child to take over the farm. For that child, I think, who grows up, we're going to provide tools to do these things better, faster, cheaper at home. Even you think about the innovations and something like tractors, you know, and you think about how much technology

has been added into these tractors and what you could do with AI, it's amazing to think about the innovations. And I mean, that's obviously one very sophisticated example, but there's lots of applications in agriculture. It's also going to create opportunities that they didn't have themselves.

So maybe they want their kids to go get a different kind of job. And AI, if we do it right, if Indonesia makes the right decisions and investments, can usher in a whole new set of job we ever thought was possible, then we ever thought was possible. Just give you example, like I asked my middle son, you know, what do you want to be when you grew up?

And at one point, he said, a YouTube influencer. And I said, you know, if you would. Take my job. Yeah, right? But imagine if when you were growing up, you said, I want to be an influencer. That word didn't even exist. And so there will be jobs that will be created that we don't even know are possible right now. And that also gives me optimism.

We can't slow down progress when it comes to technology. What we can do is make sure that people are sort of helped during the transition and given the best information they can. This is kind of what gives me hope. And that's how I would talk to anyone thinking about this. But in the end, we have to make sure humans have agency on this, right? Not that they hear it. If we further elite ties this by hearing from some professor, that's not going to be the way it works. So what you're simply saying is that the further sophistication of AI will actually require more human capital.

I do. I think human capital will be a big compliment. I hate to get wonky. But yeah, I think a lot of people are thinking about substitution. And that's understandable. We all are anxious about new technologies. anxious about new technologies, especially with my kids. But a lot of new technology has complimented workers, made them more productive. And I don't know, I mean, Gita, I've seen this in my own life

already. You know, when I work on a presentation or I prepare for this podcast, I can use AI to prepare to learn things about you and what you've talked about before, which is really useful. Scary, man. But it's helped me now I can focus more on, you know, the points I'd like to make or preparation or not being nervous to be on a podcast with so many subscribers, right? I mean, these are things that happen so I can focus on the higher value, higher margin things in my life. And I feel like for a lot of people, AI will be that kind of productivity enhancer, and that's going to be really positive. It doesn't mean that's going to be true for everybody. Part of my job is figuring out which is which, but I do see those benefits for a lot of people, including here in Indonesia.

I want to pick up on productivity. Yeah. Southeast Asia is a bit divergent. Singapore's productivity is in excess of $200,000, $200,000 per capita per year. on a purchasing power parity basis. Indonesia is about $25,000. What hope can you give to the Indonesians, the Laotians, Cambodians, Thais, Filipinos, whose productivity per capita per year is less than $100,000, that it's actually going to move up a lot faster than whatever we've seen in the last 20 years?

We've seen, I mean, tremendous growth around sort of the global south when it comes to productivity. but it's true that at the levels you're talking about, we still need to move faster and more, right? At least in Indonesia, I'll tell you a couple things that made me confident, and having been here, right? What is the youth of the population? As I said at something earlier, you know, 70% is under 40,

which makes me older than many, many people here in Indonesia, which makes you feel like an old guy coming back after so many years. But that's-you-at me, man. Yeah, but we're still young at heart. That's the thing. Look at it, right? But that youth gives me a lot of optimism. When I was at the university this morning,

I talked to a bunch of founders earlier this afternoon. These folks are building things. They're using the API. They're building companies. They're building applications. And because there's so many young people, I mean, our users tripled over the last year in Indonesia. That is wild.

And a lot of that growth, the majority of that growth has got to be young people. And so I feel like what makes me optimistic here, what makes me optimistic that growth will happen, the productivity in the private sector, is these are the kids who are going to populate the private sector as they're growing in and building their own company. So that's a reason for optimism.

Two is I think that geographically, you're in such an interesting place. I mean, this is a crossroads for the world. I mean, it's energizing to be here and see your connections to the global economy. And I feel like you're here at a place where you have the population and the strategic position to see so many different technologies from different parts of the world. And that's exciting for Indonesia, it makes me optimistic. I do think for you to see that productivity growth that you're asking to go up to those levels that you want, the private sector has got to play a key role.

This is where the adoption of the technology really matters. One factoid, the U.S. and the U.K., both were adopting IT in the late 90s. Professor Nicholas Bloom from Stanford has a great study you're aware of, I think, from spending time there. But he found that the U.S. firms got much more benefit from IT than U.K. firms. And it wasn't that the Americans had access to different technology. It was basically the same. But they adopted it quicker. They had better management practices.

And they were able to leverage that technology and gave them a lead that, in some way, was insurmountable. We have to make sure that if we're going to push for productivity increases in place like Indonesia, that firms are the tip of the spear in terms of adopting these technologies and that we're pushing both the right management practices and development of human capital to do it. This to me is the playbook. The good thing is we've seen it before, now we just need to execute. So I think, you know, I'm optimistic about Indonesia, but there is work to do here and everywhere else. I mean, I look at you and open AI as a solution for some of the structural issues with Southeast Asia.

Southeast Asia's got 8,000 universities, 19 million students. China's got only 2,600 universities, 39 million students. They're producing disproportionately much more intellectual property. There is a desire amongst many of us in Indonesia and Southeast Asia to move up the ladder, to try to catch up, you know, incrementally. Sure. How do you translate this AI narrative or open AI narrative to the common man, common women in Southeast Asia or Indonesia so that they feel a lot more hopeful about moving out the ladder? I think that it starts with education.

You know, we all want our kids to get the skills to succeed. And to me, one ingredient in that success is going to be being conversing with AI, understanding how to use it to do your job. And so if I'm a parent today in Southeast Asia, I'm hoping that. whichever school my children go to or whichever one of those many universities they aspire to go to, they can get an education that incorporates how AI is going to change the nature of work. And if they can learn how to use it to their advantage, they're going to be more successful than their parents and then they would have been otherwise. This is the hope. I also think at the same time, universities can learn from what's made universities successful

in other parts of the world. I mean, you know, as we talked about earlier, India has produced a lot of STEM grads over a million every year. You can increase. the numbers here. I think we talked about in Indonesia, maybe a couple hundred thousand each year. Three hundred thousand. India is like one point four. One point four. So, Southeast Asia is 750 to 900,000. So how do, as a percentage of population, how do we increase that percentage every year? Well, but it's not necessarily rocket science. We know how to build these STEM pipelines. We've seen it in other universities in other areas that are similar to Indonesia. And the question is, can we do that here? The other thing I think is connection between

university and business is very important. One thing that's really worked in the United States. And at Cornell, at Berkeley, the places that we both know well, they were intimately connected to the needs of business. Trying to solve problems to the research and commercialization that business was interested in. This kind of technology transfer between leading universities in the country and the leading businesses is key. And this is also something I expect Indonesia to make big strides on. That's going to be really important to unlocking opportunities for kids. So for somebody thinking about this for an ordinary Indonesian, to your point, it's got to be my kids got to get the skills at an institution that's on the cutting edge. That's the

way that he or she is going to succeed. And you get on the cutting edge by being close to practice. And I think that's what I hope to see a lot of Indonesian universities. Those 9,000 you mentioned, I hope to see those integrated more in that way. 8,000 universities, 90 million students. China only has 2,600 universities, 39 million, a lot more intellectual problems. And this is, I mean, these are big numbers to work with and a lot of opportunity. I hope, and I think this is something you're pushing for, but a lot of experimentation and learning, maybe those 8,000 universities will do, different things, take different approaches and strategies. Some will be very successful in one area.

Others might be successful in another. Others may not find success. But if we see what they're doing and learn from each other, this is a lifetime of work to be able to do this and spread best practices. I hope that people are working. I'm sure they are in Indonesia and I'd like to learn from them too. But Ronnie, what you're saying is that countries will need to produce, still need to produce more STEM products. Draw out a picture here. If you can as Open AI to code. Why would you need a product of education in STEM? Such a great question because a lot of people said, look, now that AI can write code, why would I ever major in computer science? But what I say to this is there are so many people who have their undergraduate in engineering, but they're not actually

working as an engineer. Does this mean that their undergraduate degree was not valuable? No, because your undergraduate degree is a reflection of learning how to think in a particular way. And so I feel that, ironically, despite AI being able to do some of these coding tasks and it will get better and better, it still may be wise for those who are inclined to study engineering in STEM subjects, to learn the fundamentals that are going to underlie all the advances we're going to see. Just because they might not be writing as much code at work doesn't mean that it wasn't useful. And by the way, they might write better and more code with the health of AI. So I actually think it's unlikely that we're going to see sort of less demand for STEM skills.

I think we'll need that. Different kinds of things will need to be taught. The jobs will change. but that will allow people to focus on new and exciting parts of their job. So to me, I don't see it as a contradiction yet. I'm with you. And you've also kind of alluded to the fact that critical thinking is key, right? And you can take philosophy to allow yourself to be a better critical thinker.

And what I find with new technological platforms, we're being seduced, pushed with all these 15 to 30 second contents on some platforms. that doesn't produce critical thinking capabilities. How do you fight that? Empowering AI to fight that phenomenon. I mean, this is my biggest concern as a parent is, I think there's so much potential for AI to help critical thinking,

but you have to have the base level skills to use that. So, for example, when I'm learning about a new topic, I feel like chat chit is my, it's like a super smart co-author or somebody to be a research assistant, to help me think about things, I never could have thought about my own.

And when I interact with it, I'm obviously coming from somebody who has a PhD who's written papers. I know the questions to ask in the fields that I'm more familiar with. Even in fields that I'm less familiar with, I as an academic, by training, I can ask good questions to chat GPT

and get interesting answers and delve deeper. And I think about, my goodness, I'm doing more critical thinking at my age than I ever thought I would or could. But now I'll look up for my kids. If they're using the calculator before they can do the basic math, they might not learn the fundamentals

to be able to ask the better questions later. So at least for my kids, I want to make sure that they get an education where they learn critical thinking. They learn the fundamentals. They're still working through the math proofs or reading the difficult text. This will allow them to use AI more beneficially later.

And I think as AI becomes more pervasive in education, and as a professor, I've seen it. I've seen it. You've seen it. I think the answer is, under what conditions is it going to be helpful to create more critical thinking and encourage it? And when is it going to be less helpful? And we have to align the rules of the road,

right, whether it's the schools, the parents, with those realities. So one of the things my team is very interested in, I run the economic research team at OpenAI, is doing research into when it's useful for critical thinking. It's still going to be an important, an essential human skill if we're going to be good compliments to AI and if AI is going to be aligned with us and our goals. I see phenomenon with the young generation of their lack of willingness

to sort of reinvestigate pre-existing truths or axioms, that tends to somewhat diminish the ability to philosophize, the ability to do critical thinking. That's a bit scary, no. Oh, if we have to, I hope that our tools can be a way for people to investigate and interrogate, you know, sort of the things that we think we know in all different fields. That's what makes it really exciting and interesting. But we have to teach our kids to ask those questions, to understand that, hey, this may have been true in the 60s and 70s when your textbook was written. But now, the world is unfolding in a different way.

And so I think we got to encourage our students to ask those kind of questions. I think your point about, you know, this short form content, the quick hits is very concerning. And that's why I feel like when I... Yeah, when I engage with Chachibit and I have these turns and these cycles of getting information and going deeper and deeper and deeper, I feel like, I feel pretty optimistic that I'm actually learning a lot. We have to get the base foundation right for the younger people before they can do that kind of thing. And look, they're going to learn some things I never could have learned quicker and faster. but I do think the fundamentals are going to be important.

Also say, I mean, the success of your podcast is sort of a testament to that long-form content can still be really great, right? That you have all these young folks all around Southeast Asian, around the world, I guess, like dozens of the country, right? That people are consuming long form and they're listening to it and they're learning from it. They're engaging with it. This also gives me hope, you know, that people are interested in the content, like the kind we're producing right now.

Yeah. Without mentioning the name of the person, but, you know, he's been, sort of like mentioning about the differentiation between platforms that are seeking truth versus those that are seeking political correctness. Yeah. How do you address that? You know, I, as an scholar by training, I always want to be on the side of pursuing truth. And as you think about, if I shared with you my questions for chat, GPT, very just personal,

you'll see it's often trying to get to the truth of, you know, what is going on here? You know, on the plane over here, I ask that lots of questions about Indonesia, trying to seek the truth of economic growth, what were the determinants of how it could grow in the future, what were the current challenges and opportunities. To me, it was trying to get to the bottom of this to understand, you know, what can I do as an individual working to open AI that's going to help the Indonesian economy. So for me, the pursuit of truth is really important. But how we design these models, how we train these models, who's involved in the models, that really matters, right? That matters the kinds of truth that you're going to discover.

And so that's why it's important to broaden access and get more people involved. I was really fortunate to meet a bunch of kids at the university who are developing their tools today, working on AI. And that gave me a lot of hope that the more people we get involved, particularly from Southeast Asia and Indonesia, you'll see their values, their truths reflected in the things that we're building. And that's exciting. You know, the trait between China and Southeast Asia, China is already the biggest, largest trading partner. We buy things from China because They sell cheap technologies, right? And I want to take this a little bit further into the Global South, you know, dimension.

Members of the Global South, they can only buy things cheaply. Right. And I want to take this to your philosophy of pursuing the platform on a close source and for-profit basis. How do you relate to the narrative of the Global South that's not economically as good? gifted as those in the developed economies so that your narrative actually resonates. Yeah, it's, you know, it starts with my own parents and who grew up in the global south. Right. I'm thinking about if they were in India today growing up, you know, or in Indonesia, what, what product would they adopt and how would they use it?

And so one thing you see is like, you know, the free use of chat chit is immense. You know, if you think about the hundreds of millions of users around the world, a large percentage are free users who are not paying for the platform, able to use the tools in chat, That's exciting. You also think about, and we've been public about this now, the open weights models that we're releasing. You know, really key strategic decisions, you know, I think for an organization like Open AI, wanting to be careful that the most advanced models, the ones that have the highest

capabilities, that these are the ones that are closed to make sure that we maximize safety, but other models, right, that we can release an open weights framework, which allows people to experiment and build on them, which I think is really good and serves that need you're talking about. And the last piece is, you know, being sort of flexible by country to provide what that country needs in that time, right, in that stage. So we have open AI for countries now. We're working with countries all around the world to find customized solutions that are going to work in their context. It includes the price point, the education and training required, the infrastructure they need to build.

So this is kind of what helps me understand how we might succeed in places with sort of diverse economic portfolios. I think you're right that people are going to look for value out of their tools. And in some sense, the global south is the ultimate crucible for value. People are only going to adopt this if it can improve their lives, whatever the cost is, right? And I think if we can provide tools that provide a lot of value, which is I think where we're comparatively advantaged to your point, this is where we're going to win, not just in the global south, but everywhere. This is how I think about it.

Okay. Now, from a regulatory standpoint, you brought this up. You know, we've been witnessing the last few years how regulators have been somewhat anachronistic. You know, the regulators are somewhat h higher than we are. Yeah. And those that are supposed to be regulated are age somewhat lower than we are. Yes. How do you explain this gap? And how do you address this anachronistic numb regulatory oversight that's happening, not just in the developed economies, but in my part of the world. You know, every new technology pushes the boundaries of all different parts of society, the ones that are truly disruptive and transformational.

And AI is no different. And what's happening is, you know, having worse. in government. I admire public service. I respect the people who do it. It's hard to keep up with all the new things that are happening, much less the things that are already happening in existing problems. And so when a technology like AI comes along, which is, let's just be clear, I mean, it's being adopted at a dramatic rate. So like when we've released chat GPT, we had 100 million users in two months. So look, for a government official who's trying to run a big program, deliver service, it's unreal, right? And so let's give you. And so let's give you. people, the grace, to understand this is moving very quickly.

Second thing is the capabilities of the models are also changing very fast. You know, so if you look at what the version of ChatsyvT today could do versus to what it was in 2022, I mean, it's like- November of 2022, right now. If you remember this, it's like nine and day. We're talking about something completely different. So not just as it's moving very fast. It's also changing as you're looking at under the microscope.

So we have to make sure that policymakers and other stakeholders in the public sector have the information that they need. this is where I think business can play a really important role in terms of providing the information. Here's how the models work. Here's what it's being used for. Today I got to talk a lot with Indonesians about the use cases here to explain. Here's how people are using it to learn. Here's how people are using it to code.

Here's how people are using it to make images. This kind of information is really important to share and business can help inform policymakers to keep them up to speed. I think at the end of the day, that kind of dialogue is our best hope because I've sat in those offices in Washington, D.C. I'm sure there's a similar group in Jakarta and everywhere else we've been. who are trying to figure out what's going on by reading the newspaper, reading the memos, trying to understand it, using the technology themselves,

you can only go so far. You need to talk to the experts who are the front lines. You need to calibrate and make sure you're doing what's best for the public good, not just for any single organization, but the information flow has to be there.

Otherwise, there's no hope for regulation to keep up with the speed of business and innovation. You know, you've been in the government, I've been in the government. We're seeing increasing amount of political neurosis in many parts of the world. How do you get AI to help take subjectivity out of the equation so that that new roses

dwindles? Yeah. You got to build trust with these products. I think this is what we think about so much as part of our mission. And something, you know, take seriously inside open AI is how do you build trust with the people who are using the product? Because if they don't trust what they're seeing, when they type in something that can be fairly intimate or a question they really want to know the answer to, then we're not just

going to lose a user. We're also going to reinforce this kind of dynamic that you're pointing out across the world. You're seeing this across many countries and big decline in trust in institutions. And as chatypt is being used by hundreds of millions of people, we have to make sure that we enable that trust. A couple of things really matter. I think safety in making sure that we protect people's privacy is really, really important. So you see a big focus on that with chatyPD products.

Safety is really part of our sort of mission. People take it very seriously. There's full-time teams working at opening on that because we think it's important to people. Second thing is we have to make sure that the responses you get are credible and actually responsive to the question you asked. We spend a lot of time in terms of our product team making sure that's the case because ultimately people aren't going to use it if it doesn't give them the answers that are accurate and what they want. And of course, that's improved a lot over time. And lastly, we have to make sure it doesn't reflect any particular bias ideological viewpoint or even just the fact that, that our organization's headquartered in San Francisco and there may be different views about things around the world. These are all things we think about a lot.

And I think if you lose a threat on that, it's going to undermine some of the things that we're working to build with our users. But more importantly, it will also reinforce the negative dynamic you're talking about of declining trust institutions across the world. And to your point, if technology makes that worse, then we're going in the wrong direction.

And we really have to make sure that doesn't happen. Are you hopeful that AI is going to be infused more into decision-making apparatus within a government. I mean, it's already being deployed big time in entrepreneurship, right? Very much. Yeah. How do you make sure that this thing also arrives at? I'm very hopeful. I think, look, AI is like your co-founder now. A lot of people are saying, right, when I want to generate an idea, I did some research on this and I, before AI, before AI was prevalent. And I found that most people, most founders, they never talk to someone they don't know about their

business idea. And I thought, that seems so crazy. You know, the first thing you do is talk to someone. There's lots of times people don't want to talk to someone. They're scared. They'll steal the idea. They're nervous. They don't think it's a good idea. Now you can talk to chat GPT about your business idea and it can give you good and objective advice about what to do to pursue it. It can point you to resources. It might point you to a human co-founder. But the ability to test that ideas, I think entrepreneurs in Indonesia, clearly around the world are discovering AI for that. Now the question is, how do we help it in make better decisions in the public sector? I'm already working with a couple states in the United States, Pennsylvania and my home state of North Carolina to improve government

services. I think it's a big opportunity. The human will be in the loop, the decision maker, the policymaker, or the person who's been elected to that position. But having the support in that decision with the right data to make the decision better and deliver it faster, I think there's tremendous opportunity for AI to help in that case. And governments will find the best use cases. And someone like me can help do the analysis on how much productivity it's increasing and how much government savings is producing. That's been really exciting for me to see both in North Carolina and also Pennsylvania, two of the states,

we've been working with the closest. Got it. You know, when you do a simple search on Chad GPT, it takes up about 10 to 50 times more energy than a simple search on Google. And if you do a simple AI generated or sophisticated AI generated image on SORA, that's going to require about 1,000 to 5,000 more,

times more energy than a simple search on Google. Question is about energy. Southeast Asia has got about 400,000 megawatts worth of power generation capabilities. Only two countries are defined as being modern in terms of electrification, Singapore and Brunei at about 10,000 kilowatt hour per capita. To be modern, I think you've got to be at least at 6,000.

Malaysia, 5,000, Indonesia, 1,300. do you see this as a structural limitation for the empowerment of AI because it uses so much more energy? So we definitely need to build more infrastructure not just Indonesia and South Asia but around the world because for the AI revolution to happen you're going to need data centers

with the GPUs, the kinds of things I worked on and the chips at, with memory chips and cooling equipment and things to make sure the data centers run well. And of course, you're going to be powered by energy that doesn't make our climate crisis worse. And so as you think about those things and put them all together,

then you think, okay, we're going to need a whole stack of technologies to make AI work, to train LLMs and to do inference. And actually, it's interesting, some of the math around

how much energy it uses under what conditions still needs to be done. We need a lot more analysis. I've seen lots of different numbers on this, but your point, broadly taken, that this is a energy is going to be a key part.

Data center is going to be a key part. That is true. And this is why I think you're going to see massive infrastructure investments. You're seeing some talk about this. in Indonesia to build data centers and build energy to go along with it. It is definitely going to be a rate limiting factor to sort of the amount of compute that's available in parts of the world like here and how that can be used for local applications.

I think, though, you can see a lot of interest in investment, though, in these areas. We'll talk maybe a little later, and I know it's something you've thought a lot about with FDI. Well, look at how many infrastructure funds are being raised by some of the biggest private equity and venture capital firms, particularly in private equity. And they're looking for these investments to make. Why? Because they know that we're going to need these around the world, and they're going to look at markets like Indonesia. So I'm pretty optimistic that we can take these asset classes like private equity and direct it to places like Indonesia to make sure that the capital gap you've talked about elsewhere is filled.

Now, will it be the right amount and will it be absorbed? We should talk about these details. That to me is a great opportunity. And that's one of the reasons that open AI for countries is helping countries consider the implications of investing infrastructure and energy. We're going to need it. And I think that's going to be true regardless of how AI continues to develop. The models are going to get more efficient. The chips are going to get more efficient.

We'll find better ways to deliver the energy, but you're still going to need massive investments in infrastructure. You know, Southeast Asia gets about $200 to $230 billion of FDI per year, which Singapore gets about $100, $240 billion. Indonesia, $30 to $50 billion. The others, $10 to $20 billion each. That's not enough. You know, the numbers are such that if, if, we want to move up on the electrification to at least 6,000 kilowatt hour per capita,

we're going to need about $2 to $3 trillion. We're going to have to build a terawatt, about a million megawatts. That's about $2 to $3 trillion. Yes. It's a lot of donuts, man. It's a lot of donuts. It's a lot of donuts. I mean, yeah, we want to AI ourselves to the levels that many of our friends in San Francisco

are doing it. But at the end of the day, if we don't have the energy. So this is where the, I think you raised, you know, a good point on FDI. I think Indonesia and Southeast Asia, they got to get their act together and attracting this massive amounts of liquidity that's needed. And the good news is there's a lot of liquidity. This is about $140 trillion worth of liquidity sitting around the West. A lot of dry powder or donuts, as if you prefer. And you see this.

What's really interesting is, you know, you see these private equity firms and all. other kinds of investors, sovereign wealth funds, and you've also worked on this, looking for places to deploy capital. And you see opportunities like Indonesia, a huge market, young market, adopting AI quickly, needs the infrastructure investments and the gap is significant based on the current rate of FDI and what you need. Then you think about, okay, what makes Indonesia more bankable? What makes Indonesia more investable? And I'll take something you said before, which is, look, it's not risk. That's the problem. These private equity investors are used to thinking about risk. It's uncertainty, right? And the unknown unknowns that are a little bit more.

more difficult. Right. And so AI can also maybe potentially help, you know, need to develop an investment profile that's going to attract more of this capital. Right. And if you want to go for 8% growth in these big numbers that, you know, are the envy of the world, you're going to have to fill that gap most likely with FDI.

And if you can get FDI to the numbers that you've talked about, that's going to be the lion's share of getting to that, those growth rates that you want. So this is really the key. If you think about the components of your country's GDP, you know, you would have consumer spending, right, government services, and then you're going to need investment, right? So in capital investments, this to me is a significant part. FDI is a significant part.

And the investments in infrastructure are going to be a significant part of that. That's at least one way to think about stacking enough to get you to where you need to go. I think this resonates to the global south. The ability to improve upon our preexisting world withal to translate uncertainties into risk. But the reason why Singapore gets disproportionately much more is because they have a high degree of enforce the ability of rules and regulations. My point to you is that if you want Open AI to be successful in my part of the world, I think you've got to combine the narrative with the need for your destination to be able to translate uncertainties and risk,

but also the enforceability of rules and regulations. Without which, I think you've got to have a structural limitation. No, I mean, institutions matter. the great work by last year's Nobel Prize winner, Doran Osamoglu and co-authors, you know, Why Nations Fail, right? It was one of these things.

And James Robinson and James Robinson and Simon Johnson and the whole group. But I think one of their big points is that institutions really matter. It actually goes back to that question I had around the dinner table with my parents, right? And the countries that a lot of our parents came from or grew up in, right, weren't successful right away as the places they went. And one of the reasons were differences in institutions and rules of the game. And the countries that have upgraded their institutions, and the rules of the game over the last 25 years

have generally improved their economies. And I think this is something that's been really important for making the investing environment stronger. And I do think this is something we'll take into account at Open AI. But of course, government leaders around the world are thinking about this too. And it makes me optimistic

because I think institutions, we have a playbook in other countries that have successfully built these things. They're not always easy to maintain. But I think this is exactly where we need to go. And that book sort of outlined something that I think I've learned through my career. It is another good one.

another guest you've had, another good book. But I think it's key for AI to be able to be helpful in institutionalizing decision-making. I mean, we're seeing so many episodes of decisions being personalized. Yeah, yeah. As opposed to being institutionalized. And think one of the properties of these large language models at least is it could potentially aggregate lots of different information, lots of different viewpoints, and figure out what the preferences of larger groups could be. We spent a lot of time thinking about how to get people's opinions on lots of different things.

before we make a judgment. Sometimes that's what's slowing down, for example, investments or consensus. We can use AI tools to inform some of that to help decision makers understand if we were going to take policy A or policy B, how people might react. And so there's lots of applications waiting in the public sector. That's what makes me excited. There's more than enough good ideas and many, many things I haven't thought of for all the young people and older people to try in the world. I'm pretty excited about some of the things that we can do to improve bureaucracies and reduce red tape in particular to make some of these investments happen faster. I want to talk about democracy a little bit and link it to AI. I'm in a camp that

believes that democracy cannot just be manifested in distribution of power. It has to be manifested in distribution of public goods, inclusive of health care, welfare, education, moral value, social value. How do you see AI connecting to all these dots in a good way? I think, yeah, I think AI can improve the distribution of public goods if done right. You think about one of the challenges with sort of public good distribution is access to information. If AI can make information more readily available and trustworthy and credible, that's going to help in the distribution of public goods. Second thing is being able to identify who's in need and distribute it fairly, AI can also help in this way. One of the things I'm doing with the state government in North Carolina is there's an unclaimed property fund.

This is property that actually is owed already to individuals. They just can't find out where they are and who they are. And AI can be used to triangulate how to deliver those unclaimed properties back to the rightful recipients. So it's giving taxpayers the money that they deserve. These are things that are going to make a huge difference in terms of increasing trust in government institutions, which will allow the provision of public goods to work better. The last thing I'll say about that is just the efficiency of delivery can improve. It's once you identify who's in need, once you give them the information they need,

you can get them the public good much more quickly in this way too. So if you think about sort of a health care benefit that might be part of a public good you're providing, if that's coming through a digitized system where we now know the right person's getting at the right time, that can be tremendously beneficial and a lot less waste along the way. So I'm optimistic. I think your idea is right. Democracy is a key part of the institutions, but you have to not just distribute power. You have to distribute the public goods effectively.

Otherwise, people lose faith. This is one way to kind of help us distribute those public goods more effectively. You see AI potentially taken over a position of leadership so that you can completely take out subjectivity of the equation? I don't see that. I know people talk about that. The reason is I think they're going to be use cases. At least I feel them in my own sort of view that you want humans in the loop. Even if you showed me some test that showed that, you know, my child learns better from AI than a human. I wouldn't necessarily believe it. But if you showed it to me, I would still want a human in that classroom. And one of the reasons is I think that school is more about, how you do want to test. It's about teaching you how to be human. Maybe this 20 years or 50 years from now will seem like a silly sentiment. Someone who's caught in the older age has been passed by. But I have a sense that what makes us human is in some ways appreciating other humans and what they do for us.

I think caregiving is the same way. I think, you know, AI can really help in conversation with someone who might be isolated or needs mental health support. But at the end, the human-to-human contact, if anything might be more valuable in a world where technology is pervasive. For those reasons I think when we think about leaders, when we think about decision making, judgments that affect our lives, I think we're going to want humans in the loop. But if humans are, certainly I would, but if humans are informed

by more information, making better decisions with the help of these tools, that seems to me like the highest potential. But, you know, each society will have to make its choices. But at least for me, I think humans will be in the loop on those things. You've been asked this before, but what's the typical misconception about AI? And how do you remedy that? I think the typical misconception about AI is that it's going to change everything all at once at the same rate.

You know, as an economist, what have I learned from previous technological revolutions? The future is already here. It's unevenly distributed. The William Gibson, paraphrasing the William Gibson science fiction quote. The reason I say this is look at where IT had an impact first and look at where it took longer. Some sectors are going to be much more amenable to AI being used to solve problems than others. it may take longer in health care and education where there's a huge public sector role.

There's more regulations for a lot of good reasons. And in many ways, we still might want humans in some of the key roles in a way that might be different than other sectors. So to me, I think people need to pay more attention, in my view, to the differences across industrial sectors and how AI is going to be used. And so when they think about the impact on jobs,

the impact on the economy, using a lens that's a little more nuanced than what you might read in the news now, to me, that's the biggest misconception about AI. I think we need to look at it in a more granular way that recognizes that the impact will be uneven and happen in some places, including geographies quicker than others. This is a broader question about utopia versus dystopian.

There's a lot of people out there that are quite dystopian about the future of AI. How do you address that? How do you remedy that? First, I don't seek to remedy it as a start. I think to listen to them first. I mean, this is a new and really exciting technology with uncertainty ahead. I mean, it's one of the reasons that, you know, once, if you read the news each day, you'll see some big AI story that's happening, a story about open AI or another lab, a story about a job

that's changing or a job that's being created. And that excitement is a product of a lot of uncertainty. And that's what makes it really, you know, just illuminating to be at work every day for someone like me. But I also have to empathize with people who are saying, hey, you know, I don't think this is going to turn out that well. So the first thing to do is listen to them. Because, look, despite some of the things I've done in my life, some education, worked in a couple different places, have an advantage point. I got a lot to learn, you know? And so I really mean that. Like, one of the things you learn,

the more education you do, the more experience you get, in my view, is humility to know how much you don't know. And so you might even notice during this talk, I try hard not to make like predictions that are sort of, you know, super confident because I feel like there's a lot of things we don't know. And so the first thing when I have a critic or someone who's dystopian who's dystopian what is to listen, what is making them dystop? What's making them uncertain about the future? If I could provide a factual basis or evidence to say, well, here's how AI is affecting the job market today or here's how AI is affecting that job now. Or here's how the products are really built. Or here's how we thought about that safety issue.

I provide that. That's how I think I can be productive in that conversation. When we're thinking about what the future holds and there's a lot of uncertainty, I just try to incorporate that into my own thinking and design research and work. inside Open AI that can help address that person's questions. I don't think that you can convince people any other way. We're going to come to this with different experiences. I would say whether you believe in utopia or dystopia, use the products and understand the capabilities, right? And you may, change your mind and they reinforce it. But for me, that's been the biggest lesson of being at Open AI is just using the products every day, increasing my usage and thinking, oh my goodness,

some of the things I didn't think were possible are possible. And that's made me all. ultimately a lot more knowledgeable and insightful, whether I'm talking to a utopian or a dystopian point of view. You know, I've been thinking about this the last couple of minutes. The natural tendency for somebody in the developing economy, Southeast Asia, is to see China as a technological capital allocator because it's cheap. You buy an iPhone. It costs you $1,500.

You buy an opo, $200, right? And Opo takes better pictures than iPhones, supposedly. I use an iPhone. And some of us, or many of us, look at the West as an economic capital allocator, while knowing that the West is very much a zero to one. China, I think, is much more of a one to the next digit. They just know how to embellish pre-existing innovation and make it cheaper also. How do you fit into that narrative, right?

I mean, it kind of goes back to the earlier question, but I want to just solidify this thinking so that people out there understand this more crystal clear. Open AI is a technological capability from the West. And we look at the West as being too expensive. We look at China as being much cheaper technologically. But we want to look at the West as a liquidity provider, economic capital provider. Yeah. I mean, at some point you've got to think. think about how you can actually amalgamate your technological narrative with an economic narrative,

right? It's such a stupid point. I think, like, why has the U.S. traditionally been so good at zero to one? It's because of the prevalence of risk capital, you know, and it's, you know, I was with one of your local venture capital firms, which is so impressive this morning. And what I learned there is, like, you know, as you think about venture capital around the world, it's taken longer to develop in other parts of the world compared to the U.S., right? And the U.S. was really in the forefront of venture capital for a bunch of different reasons.

We can talk about another time. But the development of VC in the U.S. and their ability to sort of invest capital in fledgling ventures, you know, two guys in the garage, right, as the prototypical story. That's a key to a lot of what we're seeing today. It's even where our own founder, Sam Altman, comes from that tradition in Ycombinator and the things he's done in the past. And so when you think about that, right, that's what's brought this generative AI revolution

to the foray, is those kinds of investments inside. some existing companies, but also in new startups and newer organizations like the ones we're at. That zero to one is the reason that we're developing your frontier technology. It's a reason why we gained 100 million users in the first two months, all those things, right? And I think if people look to the U.S. And they think about what are the institutions that made that happen? Risk capital and the ability to attract the best and brightest people, key among those reasons. To your point, though, there's more to that now for the global South and people outside of the West to take.

One is to develop their own zero to one stories. it strikes me that on the enterprise layer, the applications of AI, the tremendous opportunity in places like Indonesia, because when you saw enterprise software come out, a lot of the billion-dollar companies were built in the application layer, specific use cases, in finance, in healthcare, and retail.

These are opportunities to go zero to one in a different way that I think could exist in Indonesia. And I think, you know, lessons from the U.S. and other places could be really useful in that regard. The last thing is when it comes to AI, the foundation you're going to build on, what do you want for the product you're using? And I think that safety and trust and these issues that we're focused on are really important.

So that's the other reason I think we have a nice combination there with Open AI products. And we have to get the value proposition right regardless of where we are in the world. And our success will be sort of dictated by that. So our growth in Indonesia, I think is a testament to some of that going the right way. But I think those two things, right, for us, the zero to one mentality that got us to this point to be on the frontier. And this, you know, somewhat unusual in the history of tech, but sort of specific, and sort of prioritization and attention to safety and alignment. These are the things that kind of make it special,

make it a really fun place to work. Talk about data security. How do you address the concerns about data security? I mean, just simple answers to hard questions. Like when people say, hey, what about my data? Enterprise, for example, we don't train on your data. So it's easy when you work at a place that has very clear answers to those kind of things. Right.

So I think data security has got to be the utmost importance. because and user privacy, because think the trust point we came up before, if users can't trust you, you're not going to get their sort of their mind share. You're not going to get their usage of the product. And ultimately for something like chatubit, it's just not going to be successful. So it's a big, it's a big thing. And I think it's something that, you know, anytime that you make a mistake, you have to quickly correct it and make sure people understand what you did wrong. So I think open AI, it's something we, it's in our DNA right now. And you'll see that from a lot of public communications. Talk a little bit about the demographics of your users and also the economic strata

for users. Intuitively, I just think that most of your users are sort of like in the middle segment. Or I could be wrong. I mean, really diverse. I think what is interesting about opening up partly because when you have hundreds of millions of, you know, sort of weekly active users around the world, you know, you have a wide spot of the population. So, you know, I think other. would be focused on a particular segment like developers. We kind of have everybody, right? And I would say that by having everyone, we have all strata society, lots of different demographic cuts. I think what's interesting is, I mean, just like in Indonesia, a little less dramatic in the

general population, though, I mean, youth are a big part of this. Like, you know, if you think about the people who are going to be the definers of what's next, the young people who are defining culture, who are going to figure out where politics, economy, technology goes, they're the people who are really well represented on the platform. They're users of chat TVT. And so that makes you feel pretty good that we're doing the right things when you have the youngest folks for the trends that is using it. But because we have such a large user base, it's very, very diverse in terms of who's using it, which actually makes the models better because you have lots of different use cases, lots of different languages. That part's exciting. So I wouldn't say we're focused on one part

of user. We just have a wide user base, which is a real privilege. Any interesting, you know, development on the bottom echelon of the pyramid? Are they using it more proactively now? Definitely. You see this, you know, because with the free version, so many people can access. That's the exciting thing. I mean, if I would have told you three years ago, you'd provide this kind of intelligence for free. I mean, you'd say, oh, my goodness, like, this is magic, right? So, yeah, we forget easily. Now we see the magic that chat TVT and other tools have. You do see it. You see a lot of people using it for learning and upskilling. That's the thing that's been useful for me. So when I analyze the aggregate data, I say,

what are people using this for, right? A lot of people, particularly people who are just starting off or younger in their careers or fewer resources, they are using it to learn and upskill to get better at their job, to get a better position, to interview for a new role. And that's pretty inspiring for me. I feel like people need to take that more seriously, that by getting this kind of guidance and mentorship and advice from chat GPT, people can move up the economic ladder. And it's something I like to see and something I want to work on in my career to help the models be more effective in that regard. I don't use it that much. How do you convince me to use it more? My kids, my staff use it too much, I think, in my view.

How do you convince them to use it less? Yeah, that's a great question. For you, Geith, use voice mode. Okay, voice mode is like a revolution. I mean, if you use voice mode with the camera, you're going to find some capabilities that you're just going to be stunned by. When I go home after this trip,

my wife will have a long list of things around the house to fix. And I'm no good at those kinds of fix it jobs around the house. But with my camera and a voice activated chat, GPT, I can fix anything in my house. And I think that's the kind of wow factor that's really going to get you to use it more if that's what you want to do. For those who are using it too much in your view, I would first look like, what are they using it for? They need to spend time with the parents.

With their parents? That's interesting. Oh, that's interesting. How are you convinced them to use less? Yeah, yeah, yeah. So I would think, look, human context is going to be really, really important. I mean, as much as machines are getting intelligent and more capable, I think what makes us human, the connection has actually become more important.

I have a counterintuitive view on this because we're going to appreciate human. more because there's some things that only humans can do. So if you want to remind your kids, hey, you know, put down the phone, don't use chat GPT, talk to dad. I think that's the best way is just to remind them about that. Or, you know, you could, you could spend more time with chat GPT and train your own custom GPT to talk to your kids,

but I think that might be too far. What's your mission in Southeast Asia or in Indonesia? Oh, for us, this first visit for me is to learn the dynamics of how people are using it and how best we can support them. And when people tell us, hey, you know, the voice mode is great, but you need to improve this Indonesian dialect, we can take that back. When people say, hey, I want more support on Cody use cases or advice, we can take that back. And when we hear from the young people, the students and the founders that we met earlier today, we're thinking about how to build for them. That's like why we're here.

And so for the beginning, I think just the listening tour to understand where things are at. I think, you know, going forward, we want to grow with you. This is a dynamic place. I mean, you talk about this on your podcast all the time, but this is a place that's going places. It's undernarrated. Yeah. And it has ambition and you're trying to tell the story of that ambition.

But very much like OpenAI, I mean, our job is to tell people about this unusual, new and amazing technology that also needs a story. And I feel like those stories can grow together. It's one of the reasons we're here in the region. I'm excited to keep coming back. I want to just throw you some statistics. There's, you know, there's 400, there's 700 million people in Southeast Asia. It makes up about 9% of the global population.

there's about what 140 150 million books published of which only 300,000 books are on southeast 300,000. So population-wise, we're 9% books-wise were less than 1%. It is right there. It's very striking. It tells you we're under-narrated. But this is, see, but either this is interesting because this is telling me like, you know, why there's such demand for what you're producing here.

We talked earlier, the interesting paradox that in a world where people are looking for short-term dopamine hits. They're watching a long-term or listening to a long-term show like this. But it's because you don't have enough of the narrators. You don't have enough of the storytellers for a region that is, you know, should be punching above its weight, but it's not. I think you've identified a market gap in that way. And that's really exciting.

And I think it's interesting to tell what kind of stories are we going to tell about these days going forward. And I imagine AI is going to contribute to that, too, both visually and with voice and other tools we have. I think this region, all it will take is maybe about 20 to 30 really good storytellers to tell the stories about the regions of the world. See, this is interesting. We don't have any personalities. Let's just AI to hell out of these.

I mean, think about what you could learn from talking to chat. You can about what those stories could be and who should tell them. That's the way I would use. I would say, we're not short on stories. We're short on storytellers. But it could also maybe improve her storytelling. Maybe somebody has a story to tell, but they don't have the confidence to tell it. They don't have the position in society to tell it.

They don't have the platform to tell it. Here's where tools can help them develop the story to refine it, to figure out what about the story is compelling to other people. You know the stuff, your vast experience, you've dealt with this, but other people need that training and support. So this is where I think it's exciting. And I'm hopeful to see some of those stories, too. I think there's many of them.

When is AGI? Well, you know, people have different definitions of artificial general intelligence, but generally people think about it as, well, a large number of people think about it as the idea where sort of machines can do the vast majority of tasks as well as a human. And they often divide in a cognitive task and they make different cuts. I think at the end of the day, for me, it's about when it's the most useful, not just when it can perform particular task.

And so I think we are still in a place where AI is going to be useful for lots of different things. But the ability to do sort of the vast majority of things I do at work, we still have time before that. I think it'll be a while for most of us. I think we're already at a point, though, where it's helping people do better. at work and do their job better. I think that time has already arrived. So I expect us to keep advancing. We will keep building towards AGI. But to me, the ultimate evaluation of that is going to be

sort of are we delivering value to our users and are they using it to accomplish things that are important to them. And I see that happening already. Ron, it's been over an hour. Any final messages for some of us in Indonesia or Southeast Asia? I would just say thank you to Indonesia and the region for hosting us. It's been an amazing learning experience. I mean, for the moment, you set foot here, you get a sense of the energy and the positive vibe, this potential we're talking about. And I think I think the thing I'll take away from this, Gita, is like, I'll be looking for those stories from people here who are creating things that we never thought possible in the years to come. Thank you so much. Thank you. We got to get you, Batik. Yes. Tomorrow I'll be wearing it. Don't

worry. Thank you for having me. Friends, that was Ronnie Chatterjee, chief economist, open the eye. Thank you. Thank you.