Blueprint for an AI Future: OpenAI Chief Economist on How AI Will Shape the Global Economy

Ronnie Chatterji: Early data suggests that there's a lot of this expanding human agency, expanding human ability, expanding capabilities. Another recent paper finds that humans are actually working longer hours in some cases with AI because there's more they can get out of it. Those two things, the pace of change and the capabilities, I think are reasons why economists need to be paying attention to this.

And this is one of the reasons I joined Open ai. Get as close to it as I can. You said it right. I mean, if we don't invest in the grid and make sure that we can provide reliable. Clean, renewable energy where we can, in terms of powering AI and the data centers that make it possible, both for training and inference, we're not gonna be able to maintain our lead.

Sriram Viswanathan: Hi everyone. This is the Tech Surge Deep Tech Podcast presented by Celesta Capital each episode. We spotlight issues and voices at the intersection of emerging technologies, company building and venture investment. I am Sri Ramathan, founding managing partner at Celesta Capital. If you enjoy tech Search, now is a perfect time to hit the like and subscribe button.

And while you're at it, you can leave us a review on your favorite podcast platform. If you're just discovering us, visit tech surge podcast.com to sign up for our newsletter and check out the archive of some very interesting past episodes. So we're, uh, very pleased to welcome Ronnie Chatterjee to the podcast today.

Ronnie, you are currently the chief economist at OpenAI, but prior to this role, you had a incredibly impressive career, uh, in both in academics and in public policy and in the government. You've been a distinguished professor of business and public policy at Duke, uh, for many years. Served as economic advisor in both in the Obama administration and in the Biden administration.

And more recently, uh, you, you were helping the whole global supply chain strategy during the pandemic, uh, as well as the, you know, much talked about. Uh, chip Ack, uh, which is really a, a major, major policy initiative that would be really good to sort of understand what you think, uh, with the new administration and the effectiveness and the seriousness of that policy that the prior administration did, which was close to $50 billion plus.

And now you've stepped into something very, very exciting, which is your role, uh, at the center of AI disruption. Which we wanna talk to you about, which is your role at OpenAI. So Ronnie, it's a great pleasure for me to have this opportunity to talk to you. So welcome to the podcast.

Ronnie Chatterji: Oh, thank you for having me.

It is an honor to be here and I like these kind of conversations with people who have made a huge impact in finance and technology and building companies. It's, it's, it's so exciting for me 'cause I learn a ton from these and I just wanna say. Very familiar, obviously, with your career and your, your company and what you've done around the world.

So, um, I think we're gonna have a great time, uh, and looking forward to diving into these topics.

Sriram Viswanathan: Yeah. Wonderful. Alright, well let's, let's jump right into it because there's so much interesting, uh, uh, vectors that we can take this conversation with. As I said, you know, we have to obviously talk about your current role, but I wanna also spend a bit of time on.

The advancement of AI and adoption of ai and from a regulatory standpoint, how country should be thinking about it. And then we'll obviously touch on, as I said, uh, uh, talk about the CHIPS act and what, uh, what the country, uh, really needs to bring back, uh, semiconductor supremacy that, uh, we always had.

But I think.

High priorities of someone like Open, working on, you know.

The company brings to the market in the, in the economic terms, uh, I would think that there are probably two or three, um, you know, sort of long polls in the tent if it were. Um, one is obviously looking at the. AI safety. Mm-hmm. The whole implication of, you know, what AI can do, uh, you know, uh, is it safe, is it safe for all age groups?

Is it safe for all, all, all kinds of jobs and so on and so forth, and what it could potentially do to, you know, humanity itself. That's a big, you know, target rich, you know, uh, segment. And the second is what you said, which.

Uh, labor retraining and all of that. And the third is really the underlying debate around whether AI is more like electricity. Which is, you know, which is a utility mm-hmm. For everybody. And then you know how to regulate it, you know how to incentivize people to, you know, build more of that, uh, infrastructure and so on and so forth.

These are three kind of different things, is how I think about it. Does this make sense to you as the three most important things that someone would wanna focus on?

Ronnie Chatterji: It's a great way to think about it and I, I think that I'm squarely at least into my initial work in that middle bucket of thinking about, okay, how is this gonna affect the job market?

How is it gonna affect GDP? How are we gonna see this work through organizations in the public and private sector? But to your point, I think it's a really great way of framing it. That has implications for the other two categories you mentioned. Right? When you think about the way that, um, intelligence can be used to support workers, you're also gonna start to think about some of those safety issues.

You mentioned things that are not directly in my purview, but I spend a lot of time collaborating with colleagues across the organization. And to your other point about kind of how AI shows up in the world, how it interacts with, uh, the rule of law, our public sector institutions, to make sure that everyone gets access to this.

Because at the end day, and this was kind of interesting to me, join Open ai. Benefiting all humanity. That's the, that's the remission. People take it very seriously no matter where they are in the organization. I take it seriously. And that last piece is key to making sure that this intelligence revolution benefits all humanity.

So again, other people are working quite a bit on initiatives like Stargate or working with other countries in terms of what the rules of the game are around ai, but economics is a big part of that. So I see my role really in that middle bucket that you mentioned, branching out and supporting the work of the others in those areas.

I got it.

Sriram Viswanathan: So OpenAI had recently published this whole, uh, economic blueprint for the US and making policy recommendations. I assume that came from you or your inputs? You know, pretty

Ronnie Chatterji: significantly, I assume. I mean, a, I'll credit a large group across the organization, the Global Affairs group for, for this.

But, uh, but you know, I think a lot of this just. It's, you know, it's economics 1 0 1, updated for, uh, a new generation. But there are some differences. I, I think in the past, and this reflects some of the work I did, um, you know, in policy we sort of had this idea that, you know, the market will ultimately decide where things are located.

Let's say the production of semiconductor chips or where artificial intelligence is developed. And we really don't care where stuff is, uh, is sort of located geographically as long as we can trade and have access to it. But a world where geopolitics matters a lot more than it has for a long time. A lot of those economic theories, um, need to be revisited.

And I often say to people, econ 1 0 1 said, you know, just make the chips or train, uh, the models wherever it's cheapest. Econ 1 0 2 though, says that if that's geographically concentrated in one part of the world where you might not have holes of access to it, it can have really significant. Externalities to use the econ term right.

Impact on your national security. And that's where I think we connect some of these topics. And that's an area where I think, uh, a lot of economists are starting to spend a lot more time on. Uh, that economic blueprint is a way to think about, Hey, here's why we have to invest in AI infrastructure, uh, in countries to make sure that we have access to the cutting edge models, uh, and the inference that provides to provide intelligence for all the applications.

And that's, that's sort of one of the reasons I think that was really important.

Sriram Viswanathan: I mean, this is, uh, this, I can see exactly why, uh, you know, the folks at OpenAI would've found your background and sort of working on CHIPS Act and sort of looking at another major technology, uh, development that has national security concerns.

You know, uh, you know, TSMC, blah, blah, blah. All of that will become extremely relevant. But let's, let's gains you. Uh, AI researcher or a developer or any company that is working remotely on ai, you know, can't stop talking about all the greatest economic potential, but very seldom, or you know, or not as much.

They spend time talking about the negative impact. Uh, to really labor. So can you talk about what are the tangible gains you expect from ai, which can be labor accretive or at least labor enhanceable versus things that are labor disruptive? You know, healthcare, I can see how there's lots of interesting possibilities.

Education, there are lots of great possibilities, you know, manufacturing. Not so sure. Right. So can you, can you elaborate on where do you see labor accretion versus

Ronnie Chatterji: labor disruption with

Sriram Viswanathan: ai?

Ronnie Chatterji: It's something I've been thinking about a lot. So let's start with some of the opportunities and the positive and then we can pivot then and talk to some of the other issues which are really important and the need to be confronted as well.

On the opportunity side, you know, when you look as an economist at the history of technology. The easiest thing to say and to see is technology's often emerged as a sub, as a compliment rather than a substitute to human labor When it is introduced, and in cases where it has substituted, we've moved on to higher val margin, higher value jobs.

You start with, like you mentioned, I think electricity and the steam mentioned and the internet and the integrated circuit. You look through these different technological changes and you read what people were thinking about at the time. The idea of how they would be used and how they would enhance human sort of productivity weren't always well understood.

We had to figure it out with the integrated circuit, for example, after it was invented, before it made its way into the calculator and made its into all these different applications. When you think about ai, the first thing looking at technological history is it'll be a compliment for human labor. And if you look at the early studies, and there's now been a few peer reviewed studies coming out from economists, some sociologists, some technologists on how AI is being used at work and what individuals are.

Do eat with it. If you look at like a large consumer, uh, goods company or professional services, they're finding pretty big productivity increases for humans. 20% on order of that, in terms of the tasks they're doing, and I think this is really interesting, but also quite expected in terms of how technology's been manifested in the workforce for a long time.

I mean, you know this, but like, you know, in the turn of the century, 19 hundreds, I should say the 20th century, you know. Most of the American workforce was in agriculture, and then we had a technological revolution in that sector. And you've seen this time and time again, and now 2% of Americans work in agriculture.

But the other workers, right, the ones who aren't in that sector anymore, are working in other jobs. And our GDP has risen and we produce more food than ever before. So for an economist, I'm thinking, well, look, the early data suggests that there's a lot of this. Expanding human agency, expanding human ability, expanding capabilities.

Another recent paper finds that humans are actually working longer hours in some cases with AI because there's more they can get out of it. So that's what I'm seeing on the early studies. We'll, we'll talk in a second about whether that will persist and how to think about it. But that point of view is very consistent with economic theory, consistent with how we thought about technology and consistent with big productivity increases in the economy.

Economists, I'll say, I'll stop here, they're, they're all over the place in terms of the economic benefits, predictions, business, you know, economists, they say, you know, they always have on one hand and on. Harry Truman, the president, uh uh, he famously asked, can I just get a one handed economist who tells me one thing that's not how our, how our deal works?

I apologize to President Truman and everyone else who expected that. But look, economists have predictions that are on the lower end. You know, uh, SLU Nobel Prize winner has a more modest, uh, prediction about how much AI will affect the economy. You see other estimates that are much larger, but even the lower end estimates, um, are still quite significant in terms of, you know, tens of billions and hundreds of billions of dollars of economic activity.

And so I think you're starting to see the seeds of that, uh, in how it's being applied across the economy. And I expect it to be a big contributor to economic growth going forward. It'll probably be unevenly distributed in terms of which countries, which industries, and that's been true of every kind of technology.

That's what we've been seeing so far with the peer reviewed research and also the economic forecast. I'll stop there just to see if you have a, a follow up and I can talk about some other time. Well,

Sriram Viswanathan: I, I, I think, I think you bring up a very good point. Uh, it's good to hear from an economist that this is really.

Stages of the evolution of this technology. And nobody has the clear answer to, you know, the true implications of this. Um, but there are some very, you know, the pace at which things have gone about is what is scary about this transition. Yes. And, and in fact, um, you know, you have young kids, you have much younger kids than I do.

But you know, I have a 30-year-old and he's obviously, he's in the, in the working, uh, yeah. Uh, age. But if there was somebody in high school or in college today, I'm hard pressed to tell them what they should focus on. Because if you, if you read what meta. Puts out or even open, AI puts out, or Google puts out, you know, they expect 70 to 80 to 90% of all the code.

That written by these companies to be written by ai. So you know, why do you need a hundred thousand people in meta anymore? Why do you need a hundred thousand people in Google anymore? This is not, you know, 10 years out problem. This is the next three year out problem. So what do those people do and you know, are we equipped to be able to sort of retrain these people and are there enough jobs for them to do other things that are AI supplement able and a creative to the overall, you know, value creation?

How do you think about that?

Ronnie Chatterji: It's, it's a great question and I, I do starve from my own kids, like you said, and when they think of, Hey, what do I wanna be when I grow up? And, um, a couple things. And I, I think your point is well taken. Like economists, we think about this in terms of the data we have in front of us and, um, and technological history and what we've learned from those other studies.

What is unique about ai, and I think I try to bring this up in every talk, and I learned this from open AI and research, is open ai. Is different in many dimensions and, and one of those differences is speed. You know, chat, PT got to a hundred million users. In two months. And if I compare this to all my wonderful examples that I'll show, uh, researchers here and other people about the integrated circuit or electricity or the steam engine, you know, that's a lot faster.

And there still will be an adjustment period like this in-between times as, uh, professor OJ Agarwal Toronto calls it, that in between times will still exist before AI gets put into that killer app. Like something like Chat GPT, but really, you know. That's a really rapid pace. The second thing that's different is the capabilities of the model, and that's what you referenced.

I mean, if you look at the benchmarks, um, that we are saturating with these models and how quickly we are achieving those goals, it's moving much faster than it was five or 10 years ago. Those two things, the pace of change and the capabilities, I think are reasons why economists need to be paying attention to this.

And this is one of the reasons I joined Open AI to get as close to it as I can, and that is creating. A lot of uncertainty, excitement for some anxiety for others about what the job market's gonna look like this year, next year, and otherwise. For my kids. I think the other thing, and if people who have college aged kids will know this well, there's other macro and structural economic factors that have made the job market more difficult.

Like the internship market wasn't that great last year and there's a lot of uncertainty this year for kids who are thinking about internships and you know, people have it quit their jobs as quickly. So the hiring has kind of slowed down separate from ai. So I would say there's lots of things going on that create that anxiety, including fears about new technologies.

When I talk to my kids, I say, look, you know, when my, when I was young, my parents just gave me really a few options of what to do with my life. Um, and my parents made me more, you know, the typical immigrant story. There's two choices, like right, like, you know, either medicine or engineering, and if you're creative, you can become a biomedical engineer, right?

Something like this. My parents were, were more open than that and I became an economist, but still, there was a limited number of things that seemed practical and right. That was based on their projection of the future. I'll say. Medicine changed a lot from when I was a kid to where it's now. Engineering changed a lot from when I was a kid to where it's now.

So the idea that we could ever precisely point and say, this is what you should do, it's the old Dustin Hoffman. We're old enough to talk about this in the, in the graduate where he says plastics, it's gonna be plastics. That idea, I, I would say it's a little bit of looking through the past. At the past, through rose colored lenses, we, we didn't have perfect precision even then.

Second, I think a lot of the work that gets created is gonna be stuff we don't even, can't even describe. If I told you your kid would be an influencer, you know, 10 years ago you'd say, Ronnie, what is that? Define that. And that's, we find that in the data, like a lot of stuff that was in the census in 1940, you know, they were, we have jobs today that don't even, didn't even exist.

We didn't know what to call them. Right. So I think we should also expect, as humans always have, particularly in the United States of America, we're gonna create new work, new categories, which we might not be able to know exactly. Given all that, how do you prepare your kids? There's a couple things. I think critical thinking is still gonna be important.

I've seen this technology up close. I'm a professor of course too, so I've thought about it in the classroom. I still think critical thinking is gonna be such an important part of how we interact with intelligence and those who are developing their skills to be the best compliment to ai. Are gonna be really, really successful.

People often joke, uh, you know, is AI gonna take my job? And I kind of joke back, I say, well, no. Someone using AI better than you is more likely to do that than ai. I think there's a lot to that critical thinking and complimenting ai. Second thing I'll say is I still think financial, uh, sort of numeracy and being able to be good with numbers will be really important.

We don't introduce the calculator in math education until later. Kids learn the fundamentals and they still should, and they still might wanna major in engineering subjects or learn how to code. Not because they're necessarily gonna do that at their job the majority of the time, but there are many, many engineers who now are CEOs of major companies not doing engineering day to day.

That way of thinking, that skillset, whether it's engineering or philosophy, can be really useful even if you're not actually doing that in your job. And so I, I would encourage my kids to focus on those skills like that push forward, and they're gonna have to develop a resilience and a neuroplasticity to be ready for.

All the changes that are gonna happen, and I'm gonna try my best to guide them, but a lot of it's gonna be things that even I can't expect.

Sriram Viswanathan: Well, you know,

at Open. Who said that? Uh, you know, let me look at the various prompts. People have, you know, uh, put in for GPT and they can guess the age of the person that's, you know, putting those prompts in. Because as you said, you know, people that are more tuned to sort of understanding how to use the tool or inherently have a, you know, bigger advantage.

But, but I want to sort of talk about this, uh, in your report, you, you talk about, uh, this notion of an eco economic zone. Uh, where there are these regional AI hubs and, you know, someone once said that the future's already here. It's just unevenly distributed. Mm-hmm. And that is so true. If, if you brought somebody from, I don't know, you know, South Dakota or some, you know, inner parts of India, whatever, and they were in, uh, the San Francisco, they would go crazy about, you know, what are these cars, self-driving and all of that.

Right. If there's an economic zone that is not as tuned to adoption of high-end technology, how do you ensure that this technology is accessible to them? And is that something, one of the key areas that the economic impact that AI could have? Get greatly benefited by the, you know, what you, what you do with these

Ronnie Chatterji: zones.

I mean, first of all, I'm from North Carolina and I still get wowed by self-driving cars. So I'm, I'm with you on this. And the future is here. It's already here. Unevenly distributed. I think that, I mean, this is an animating feature of the entire organization and it includes the economic research team and, you know, someone like me, one of the advantages of being a slightly older guy in the company, I should joke, like talking mostly to 20 year olds, uh, around, around the office is.

I've seen this movie before, right? I grew up in the age of the internet. I remember having like a dial up, uh, internet connection with American online and, you know, and I saw the rise of Google search and, you know, learned how to use Microsoft Office products. I've seen all that. And I remember during those technological transformations, there was a lot of talk about digital divide, that was an old word that we used to use, referring to like the gap in internet access.

And then we talked about sort of access to high speed broadband as being like another divide. Each Techn technological revolution opened up these divides that in some ways exacerbated inequality, depending on, you know, which study you look at and how you think about it. And there's this conviction here that we, we cannot let that happen again.

One, because we think that AI is gonna have such a big impact on the way people live and work. So it's unevenly distributed. It can exacerbate an existing sort of inequalities and income and wealth, which are serious here in the United States, but also around the world. Second thing thing is, yeah, but whatcha what?

What are you

Sriram Viswanathan: pro, if I may, what? What are you really proposing that the government, what is the role of the government or the state governments to actually not have this sort of, you know, peaks and valleys of adoption, if you will, and have a more. You know, uniform sort of

Ronnie Chatterji: adoption of this technology, what do they do?

It's, it's gotta start with adoption in the public sector and in educational institutions. You know, when you think about it, like, you know, I went to public school growing up. What gave me the opportunity to use different software, different hardware. It was available in my school and I great teachers who were building lessons plans around it and working with it.

I got to use it. And at the extreme level, you know, the Malcolm Gladwell's book about like the 10,000 hours. Bill Gates got a chance, right to live near a center where they had a really high performing computer and he got to put in the time we have to make sure that people have access to actually use these tools.

And you're seeing like, you know, the government of Estonia just announced like a big deal to use AI across the economy. They did the same thing back with the internet. So I think the role of the government is to make sure that. Both in the public sector and in the education system and hopefully among their firms in their country, that you're seeing more adoption and experimentation with these tools.

If you don't see that, you're more likely, I think to see a have and have not situation where the companies and the countries and the institutions that are getting to know this, they're doing the prompt engineering. You're talking about getting familiar with it, unlocking things they couldn't do before.

They're gonna get better and better, faster and faster driving those divides. And so that's where I see it to be really, really important. How you do it, how you do the rollouts, how you incorporate it. That's gotta be the decision of, you know, political leaders and other folks. But I would say making sure your people are trying this stuff very important.

And I, you know, I saw that as we talked about before, I saw that when we went to India, just how many people are using it in India and what a priority is. For the government to get more people to use it.

Sriram Viswanathan: Yeah. I, I wanna get to India and China in a bit. Yeah. But before we do that, you know, in the old days when broadband penetration was very low, there was this huge initiative to sort of push for greater broadband adoption.

There were lots of, you know, the, the famous Telecom Reform Act was there and uniform access and all of that. So you've actually called for a national transmission. To fast track transmission of, you know, infrastructure, including ai. How critical is upgrading our energy and internet bro backbone important for, for AI growth?

And, and so help us understand, you know, how, where are we in that continuum, if you will, of uh, let's say five years of, uh, of infrastructure development. Where are we in that journey to get uniform access for ai?

Ronnie Chatterji: I think we're at a critical phase. Like it's, it has to start now. And I think you've said it right.

I mean, if we don't invest in the grid and make sure that we can provide reliable, clean, renewable energy where we can, in terms of powering AI and the data centers that make it possible, but for training and inference, we're not gonna be able to maintain our lead. I mean, you talked about it, you know, San Francisco is, this is this amazing place for ai.

There's like 800 startups in ai. There's companies like Open ai, um, of course the United States is benefiting a lot from that. But we will only be able to keep that lead and keep right these organizations vibrant if we invest right in these assets. And it has to be, to your point about benefiting all of humanity and, and sort of distributing the benefits across the country.

And the great thing is there's different places that have different attributes to make that, uh, possible. As you're seeing different states step up and, and embrace that. I think it's important. I think it's the critical phase. I think it has to happen now. And the good thing about it too, I mean speaking of someone who worked in government is like I see a lot of bipartisan agreement about investments in infrastructure.

One of the nice things, you know, having worked on the CHIPS Act, that was the kind of investment, uh, that people said, look, this makes sense. Whether I'm a Republican or a Democrat, whether I'm interested in politics or not, making more chips in the United States, making sure we can make stuff again and making sure those fabs have the right energy and the right workforce.

It just seems to make economic sense. I sort of see like these AI infrastructure investments as similar. I. To your point, probably one of the reasons that like this, this opportunity made a lot of sense for me because I was already looking at the world through this lens where I felt like the United States needed to make these big investments.

And I think AI is just part of that stack that we started building back when I was in government service. Yeah, so, so we,

Sriram Viswanathan: we should. Pivot to the whole geopolitics since you alluded to it, you just came back from India as well. You know, OpenAI has submitted a proposal to the government for the US AI action plan and a bunch of recommendations.

And there are lots of concerns about China's ascent, uh, in ai. And people have seen the sort of the semiconductor movie and they don't like the ending of being dependent on one tiny island, you know, one large company. Um. So, so how do you, how does one react if you are recommending to the US government, with your, you know, economist hat and, you know, being part of Open ai, CCP is determined to overtake.

The US in AI by 2030. So can you talk about that and some of China's strategic advantages to be able to get there and what is the US need to do, uh, in response or in anticipation of that?

Ronnie Chatterji: Yeah, I'll talk about it just based on my expertise as an economist. I mean, there are so many experts, uh, more versed than me on the geopolitics of it all, and national competition.

A personal reflection is like when I learned economics, you know, I immediately got interested in the economics of innovation. And I thought, wow, nothing more important than innovation. And I recognized in my training that like innovation is why countries grow. Um, and investing in the people and the infrastructure to make innovations in drug discovery or material sciences or AI is really one of the reasons the United States is where it is today.

And I wanted to understand that process. But in those discussions, you know, when I was in grad school at Berkeley, I. Geopolitics was kind of in the background. We lived in a very stable geopolitical, um, context, and at least in terms of, you know, great power competition between countries. The US was the sort of the undisputed economic national security powerhouse during most of the time that I was studying economics.

Now what you see, we're entering an era where, um, there's more sort of. Great power rivalry on the horizon. And also other nations, like you might call them swing states. India is one of them. Uh, the U-A-E-K-S-A, you know, Nigeria, all these other countries, right? And, and of course many European countries in the UK are coming out and saying, okay, what is our role in this new world?

And that creates a real national imperative in my personal view in the United States, around economic competitiveness. And so as an economist, I can speak and say, look. The reason I think the CHIPS Act and now sort of investing AI infrastructure makes sense is because the United States is gonna wanna make sure for this transformative technology that's going to really shape our economic futures globally, that the United States has an important role and a leading role.

And the way we do that is have the best AI researchers at places like Open ai. You also have to have the compute, you also have to have the energy. You also have to have the ecosystem and the right kind of education system. All those things need to be in place. And so, at least from an economics point of view, I think we need to be thinking about it strategically and making sure the US is the leader, um, as that relates to our relationships with other countries and how this work, and that's.

Probably beyond a geopolitical strategist or someone like me. But, um, but in general, I think that makes good sense for the United States to invest in those things. And, you know, we're gonna have to work together with partners and allies to, to achieve those goals too. And that's, that's a really important, uh, job for folks, uh, in government and out to

Sriram Viswanathan: that, that, that's kind of what I'm, I'm trying to get to because, you know, if you were to compare what's happening in AI with, let's say the transportation industry or the communication industry, you.

Cross collaboration between various countries was essential for reaping the full benefits of those two, you know, huge innovations. And today you can jump on a plane and you know, go from San Francisco to wherever and you know, any number of countries that you cross. There is a common system to actually manage.

Yeah, safety and the transportation. Uh, much like what you have to do when you make a phone call. You know, there is some organization that actually makes sure your call actually is, you know, there's roaming and settlement and all of that. Mm-hmm. Is there a model like that or do you see a need for something like that for ai?

Because today it feels like, you know, every country is spending gazillion dollars in.

Uh, coordinated effort, not just on AI development, but AI usage across multiple countries. Do you see a need for that? And if so, is that a model that you think that one might follow to ensure that maximum adopt, you know, adoption of

Ronnie Chatterji: this? I think number one, we do need collaboration, uh, across the entire world.

And I'll use the analogy with climate. You know, when we face global challenges, we need to work together, uh, partners, allies, and rivals. To make sure that we can get to where we need to go. And I think with ai, uh, that's also true. And if you just look, look at you and me and our background, look at the researchers at Open ai, it also means that having people from all over the world, or heritage that comes from all over the world, it's really important to coming out to like the best ideas.

The one thing I'll say though is I think sometimes we paint it in a very binary way. People say, oh, well. If we're doing this, uh, in terms of sort of big building economic competitiveness, we're not doing the other thing. I think actually we can collaborate and compete on different dimensions. It's harder, but um, you know, when I worked in government, we were at times right, obviously competing, but we also kept, and this the same is true as this, in this administration, we kept dialogues open with, with everyone around the world, right?

And that's really important. And even let's say in an era, um, of relationships between the US and the Soviet Union. You had track two and other kinds of dialogues that were happening to make sure that we were talking about the future of really important technologies. And I think, you know, while that isn't perfect, right, we do have models from those prior eras and we're gonna have to build new ones to make sure we can collaborate even when we might be disagreeing on other issues.

So I actually feel pretty confident, especially given sort of the capabilities of our governments and also the vitality of our private sector, I should say, in civil society to be able to have two conversations at once. I don't think it needs to be an either or. I think that when you talk to researchers at a place like Open Eye, this idea of like collaboration, be really important for safety.

Resonates quite a bit, and as an economist, I, I see the same, same things. We, we gotta get it right. It's not gonna be easy. There's no silver bullet or script, but we have to be able to do both if we're gonna succeed. Yeah. But that,

Sriram Viswanathan: that actually brings up two, two important questions. I think the EU has created this most comprehensive framework for AI regulation, you know, what are your thoughts on that?

I mean, is this, is it helpful to the overall. You know, case for ai or is it, is it going to, you know, slow it down or what, what's your, what's your thought on what EU is trying to do?

Ronnie Chatterji: Yeah, I mean, look, I coming back from London, Paris, Brussels, and I'll go back. I mean, um, I think I see two things. One, policymakers there are really trying to make, um, a concerted effort, uh, in the EU and the UK to understand how AI is gonna affect their economy.

And so I think they're very conscious of the idea of like. This is gonna be a big deal and they wanna make sure that they're not left behind in those areas and they have a lot of talent, uh, and a lot of important assets that really stay high. So they wanna make sure that those are leveraged and utilized.

But they also wanna make sure that the regulatory environment matches their values and, and at least their citizens, and they're trying to get that balance right. Um, the worry of course is that like if we continue to develop these, these models and advance the applications in other places, but it's not happening in some parts of the world because of regulation, uh, that those places will be left behind and they're trying to get that balance right.

We try to work constructively with them most. There are many other folks on the team who are really deep on this, but for me, I just try to say, look. Here's the impact of AI and the economy. Here's what it can happen. Here's, here's how it can affect your firms. Here's what I think adoption matters and each country, they're gonna have to decide what kind of framework they want, what rules of the game they want to, to make that happen.

I do think you're seeing tremendous growth in the United States right now. We we're the home of sort of the, the major AI labs. You're seeing all these startups. There's a reason for that. We created an environment in the United States of America over many generations. Republican, Democrat, many administrations that's really supportive of innovation and entrepreneurship and uh, I think we're reaping the benefits of that.

So that's, that's one model. There are surely others, but that's what I think is, is really interesting for me as I have these conversations.

Sriram Viswanathan: Yeah. Well that, that's great that you brought it back to the US because from an economist perspective. You, you are absolutely right. Most of the leading AI labs and AI work is actually happening in the us.

China has lot of, you know, uh, stuff going on and some of it is known, some of it is not known. But in the US is there a correct balance between regulation and innovation and AI and. You know, do you think having 700 different state AI laws being introduced right now is the right way? Or do you think there's a need for, and and, and a push towards a national or international AI framework?

I mean, what's the model? I mean, everybody, it's kind of like, you know, let chaos reign and then in the chaos, is that the approach that the government folks are taking right now, or what do you think this headed? I

Ronnie Chatterji: mean, look, from an economics point of view, I can tell you like, you know, having clear sets of rules to follow is really, really important.

Um, and, uh, and so, but I also think when new technologies come about, everyone's trying to understand them and participate in shaping the rules. I think like the one cool thing about OpenAI, and as the economists, I've just like learned this like. Mostly people here are just focused on building amazing products.

Like you don't build chat GBT unless you know people like you. Don't build a product like that unless you're really focused on like, what do people wanna use this for? What are the fun, cool things? I'm sure like you use it, you're seeing something new every day that you can do. So I think regardless, these are all important issues, but I think in the end, the, the company's very focused on research, um, and sort of making an awesome product.

I do expect, like, as AI develops for sort of the rules of the game to be clarified, for us to have, we've done this in the United States before on so many different things, so I'm pretty confident, I think actually like letting policy makers work through their process, constructively, working with 'em. We have so many people at Open AI who are doing things like that, that to me, that's the best, that's the best way to go forward and I'm, I'm very confident when it comes to the US just given I have context here.

I've worked in government. We'll get to a good, a good place because we've done it with technology in the past and managed to lead the world. Um, that's, that's how I'm seeing things, you know, and I think that the more constructive we can be great. And, and, and, and, but we, we really do have to focus on building amazing products is really, everything depends on that.

Sriram Viswanathan: So, so let's, uh, let's get back to the last, uh, big thing that you worked on, which is the CHIPS Act. Yeah. I mean, it's, you know, you, you've got a new job, there's a new administration. Yeah. The market dynamic seems to have changed quite dramatically. Um, you know, some would say the CHIPS Act was great in its vision.

Um, and strategy, but sort of bad in execution because that's just the nature of the beast. Mm-hmm. You know, administrations change and not everybody is bought into it. So before we get into the specifics, if you were to just step back and look at CHIPS Act, you know, the vision and strategy, the actual implementation, so to date mm-hmm.

And where is it headed in the future? Can you just elaborate? You know, how do you see it?

Ronnie Chatterji: Yeah, vision and strategy. You know, I was one of the people who worked on like a strategy document outlining like what our goals were gonna be. Um, and I take a lot of pride in that work that was done, you know, led by Secretary Gina Raimondo and the chips office at the Commerce department.

And my participation for the White House at the National Economic Council with people like Brian Deese and Low brainer, um, you know, we tried to put out sort of a strategy. That would give some goalposts that wouldn't be moved. I think one of the big challenges when you make these big investments, it's often called industrial policy around the world.

You know, government makes a bunch of investments, they change the goalpost. Uh, people say, well, you didn't achieve your goals. And you say, well, no, they were different goals. You know, I'm, I'm cutting up with something new. We laid out what we wanna see in the United States with leading edge clusters to make the most advanced chips with r and d with the role of memory, um, you know, sort of with legacy and mature nodes.

And so all those things were written down and I think that was a really good thing 'cause it kind of gave people a playbook and gave market participants an understanding. I think it's really important. In the work we did, we also were, the timing I think, you know, was important. Ai, the AI revolution was.

Was kind of taking off in the public imagination around this time. And so, you know, chips, right GPU and memory chips are gonna be really, really important in ai. And we kind of saw that and uh, it was good that we did the chips act when we did, given what happened, uh, with ai. And I think that's one of the reasons you had bipartisan support for it.

It's something I'm really excited about. You know, I know we live in times where people have. Big disagreements about politics, and that's really important part of living in, in a democracy. I participate in that myself, but like, this is something that there was just a lot of bipartisan support because it just makes sense.

And so I like the vision, the strategy, bringing people together around what we were gonna do with implementation. We always knew it was going to be, uh, challenging because this is like a, you know, a live market where you're bringing in, uh, $52 billion to build these huge facilities that take. Years to put together and bring the equipment in a workforce that hasn't been at this scale for a long time.

And so, um, those things, um, I think it's, it's harder to predict. It also relies on like the fortunes of companies, the ones involved in the industry, uh, TSMC, Intel, Samsung. I think what you're seeing though is like, you know, t SMCs projects in Arizona, Samsung's projects in Texas. RD investments, uh, around the country, uh, ranging from Indiana to Oregon and a lot of places in between.

You're seeing a lot of, uh, big investments in semiconductors across the value chain. I think that's really good for the United States of America. Um, as we transition from one administration to the other, I mean, you know, that's always, when you work in government, you know, that it, you know, you might be working on something, but then tomorrow someone else might be working on something and they might have a different viewpoint on it.

But I think what you're seeing is, um, again, the bipartisan ethos of this is carrying forward. 'cause people really do care. About making sure chips, uh, are made in the us they might use different tools. I imagine you've seen lots of discussions and we'll see how it actually shapes up. I don't have any insight to that, but I feel pretty good about like, um, the decisions that we've made along the way and where we're going.

And I think more broadly, the vision seems to be this is really important and how we get there is gonna be determined by each administration and how they implement it. But overall, feeling good about it. I think that, um. You're not gonna be able to replicate an entire supply chain in a country overnight.

And so I never had this expectation that that was gonna happen very quickly. I do think it's really changed the conversation around chips, uh, which is very important.

Sriram Viswanathan: I, I think that's, that's a very fair way of, uh, uh, representing what's happened to the CHIPS Act, and I completely agree with you that the vision and the strategy was well thought out, but, you know, if you were to compare that, uh, to what's going on in ai Yeah.

Uh, you know, Microsoft announces, you know, $80 billion investment and open ai. You know, 50 billion is a lot of money, but it feels like in the AI landscape it's Tuesday.

Ronnie Chatterji: Yeah, no, that's, but you know, even in the, in the semiconductor fab landscape, it's a lot. It's, it was money. But you know, if you think about the CapEx of the economy, even that, right.

And now ai, but I, but I actually see the AI investment boom and this infrastructure boom has building on top of right, those investments from chips, meaning. You are always going to have to leverage that $52 billion into a larger amount of capital investment from the private sector. And that happened like around when I was there, like around $300 billion of private sector in, uh, investments were announced.

So I kind of feel like these subsequent investment booms, I mean they're not all 'cause of the CHIPS act, I wouldn't say that, but it's, uh, it's complimentary to, to those investments.

Sriram Viswanathan: But that's a slight difference though, I think in the way that. You know, government's role of providing incentives, you know, whether it's solar infrastructure or EV or OR CHIPS act, or any of this is where private capital doesn't naturally find it attractive enough to go into that large patient capital kind of an approach, and the government provides the incentive in ai.

It feels like everybody. Is putting money into ai. And some would argue that it's, you know, heavily, heavily hyped up and yeah, overvalued and all of that, but there's no dearth of capital. So given that mm-hmm. Do you think that there is a case to be made for some sort of a, you know, a sovereign push towards build out of AI infrastructure backed by public funds, or do you not see that?

Ronnie Chatterji: I think so. I think you're making an important decision, like when you think about the CHIPS Act or other policies like that. The role of government, I think we wrote in an agreement is to lay the foundation. That amount of money isn't going to solve the investment gap between the United States and other countries.

It's more of a carrot, it's an incentive. It was a carrot, and it comes with a tax credit. Of course, that was kind of creating more investment, all really, really important things. There's also investments in r and d, which is something that might not be funded as much by the private sector. Left to their own devices.

When it comes to ai, you're right, you're seeing massive private sector investment. Whether it is enough though, to provide the compute we're gonna need for the future and whether that compute is going to be distributed enough to benefit all Americans, that's a good question and that's where I think I.

You know, government will play a increasingly important role coming forward. So you are seeing Massives investment numbers, and I think you're right to say, you know, what's the right amount of investment? This is how markets work. But government's gonna have a, an important role, I think, to lay the foundation, to make sure those investments, um, are working, are they're in the right places and they're benefiting people.

And also, to your point about the grid, you're gonna need upgrades to the grid no matter what, to support all this investment as well. So those are, I do think you'll see a similar complimentary relationship going forward, just the same way we envisioned when we think about something like the CHIPS Act.

Now you think about it in the case of ai. But pleased to see all this investment and and excited to see where it goes. Exactly. Yes, I agree with you. Those are big numbers. Incomprehensible to me.

Sriram Viswanathan: Yes, that's exactly right. Um, you know, you've traveled a lot in your current role and even before, uh, you know, you've seen in firsthand what's going on in China, you've seen what's going on in India.

What are you excited about these two geographies as it relates to ai? And is there something we in the, from those geographies,

Ronnie Chatterji: I enthusiasm of the. Populations. I mean, I'll speak for India. I just got back from there. Right. I, I haven't, uh, haven't, you know, visited that many other parts of Asia yet in my, in my travels at OpenAI.

But in India, I mean, gosh, I, I, I might get the number wrong, but I think it's like a million STEM graduates coming out every year. Something. So when you see,

Sriram Viswanathan: what did you say? I said 1.4 million one.

Ronnie Chatterji: I see. You know, so like 1.4. And I felt like at this talk, um, you know, maybe half of them were in line and we were talking 'cause it was like amazing to see how many people were there and every kid.

Was so excited about ai, and to be honest, I can say this to you. I mean, they knew more about AI than I do. These kids are super teched up super smart, asking me all these questions, like, wow. And I can only imagine what that 1.4 million looks like at scale. And so I think for, for Prime Minister Modi and the government, they're thinking, oh gosh, we have all this talent.

How are we gonna deploy them in the right way? And, you know, look, looking at the enterprise software revolution of, of, of a decade or so ago, or longer. A lot of the billion dollar companies were built at the application layer, and I see a lot of Indians really interested in building applications on top of APIs like ours, doing amazing stuff.

There's obviously also a lot of interest in building their own large language models, but I think what I'm seeing a lot of action, at least in terms of venture capital funding companies started is on the application verticals. And you know, we both know India knows how to do that, right? They've proven that in the last generation of startups and companies they've scaled.

So that was pretty much really exciting from the India trip that the, the energy of the youth. How switched on they were about AI and their dedication to pursue their interest through entrepreneurship and innovation. I mean, that's just, it's really good for the world. Obviously, it's really good for India.

I think you'll see a lot of great companies come out there that are gonna benefit people, uh, all over the world and, and transform India's economy. If. If they get it right. So I'm, I, that was pretty exciting for my global trip.

Sriram Viswanathan: So, is there, is there anything that, the second part of my question, is there anything that we can learn or do differently based on what you've observed in India?

Ronnie Chatterji: Yeah, I think the, the optimism is really fun to see. You know, I, I'm in a role right now where I hear a lot of excitement, but also a lot of anxiety. I think the excitement is often coming from young folks in other parts of the world who are saying, oh my gosh, this technology is like a leveler. You know, this is gonna let me do things I could not do before.

And it's important not to lose that optimism. I, I also think there's anxiety. The job market's changing in ways people don't understand. I think about it for my own kids. I get that. The optimism though about like this could be, um, something that could be good for us and good for opportunities is, is infectious and I think that we can learn from it.

I also think we, we need to think about the, the negative consequence. We need to track them, we need to figure out indicators. I don't wanna. Paper that over, but that I think we can learn from, from that optimism, excitement about, hey, how can I build some cool products that could do things that we'd never done before?

You, you feel that vibe in India when, when you talk to young and Indian, graduating from from college right now.

Sriram Viswanathan: That's great. Alright. Well, you know, Ronnie, you are, you are terrific. I think you give such a broad perspective on all the things that one has to be thinking about from a policy standpoint, from, you know, the economic impact and, and labor and all of that.

So in the last section, let's, let's just do a, you know, fun. Sure. Uh, you know, quick, you know, lightning round of, uh, short question perspective from you. So, so let's go here. Um, what's. Biggest misconception about AI today?

Ronnie Chatterji: Oh, I think the biggest misconception is that it doesn't compliment human flourishing and productivity.

I think there's lots of coolness I think I'm using in my own life in ways that compliment me and, and I what I mean by that in the economics terms, not that it gives me compliments. Since we're not doing a captions here, I should be clear, right? Compliment, meaning that it makes me better the things I'm doing.

Like I, I feel like when I give a presentation now. I've made the slides using chat, GBTI can focus on the voiceover, the interaction with the audience. Those in my business are higher margin, higher value things, and I think that the ability for AI to do that, people seeing that and, and that opportunity, that would be the biggest misconception that I think people might be missing that um, right now.

Sriram Viswanathan: To go back to your earlier point about what you might advise your, uh, young children going into college about the need for critical thinking. Are we creating a generation of prompt engineers versus critical thinkers?

Ronnie Chatterji: If we do not thoughtfully design the way we teach kids how to use AI and the classroom experience, we could definitely risk in that direction, but I, I think we can do better.

You know, we, we integrated, um, hardware and software into the classroom before. Or, and there's like mixed success and, and failures along the way. Let's be honest, I see this with my own kids in screen time. Okay. But like we got, we gotta be thoughtful how we design this and, and America has a, America at least has a pretty decentralized education stem.

So you'll see state and local, I think taking a big role here to figuring this out. So no, don't want that to happen. We need to teach kids how to do critical thinking. But I think the right kind of critical thinking interact with massive intelligence at a low cost. Oh my gosh. Sky's the limit, but we gotta get that right.

Sriram Viswanathan: I'm with you. Alright, let's, let's keep going. So what is the most important economic indicator indicators or related to AI that you check, uh, every morning, every week, whatever. Yeah. Is that, is that a metric?

Ronnie Chatterji: Uh, there's not one metric out there right now that says, here's the state of the AI economy. What I do look at is.

I look at the markets, I try to understand because it's particularly sort of the global markets. A lot of 'em are driven by AI right now. I tend to look how they're performing and often it's, uh, coming down in terms of the composition to a few key stocks that are really sort of, in many ways AI plays. And so I think that's one way to look at the healthy AI economy.

It's a pretty coarse way. And then I do look at sort of job separations in terms of, you know, people jumping from one job to another. As well as like sort of the unemployment statistics in the US and around the world, uh, monthly or whatever cadence they would come out in other parts of the world to, uh, to kind of see how things are going in terms of potential impacts from ai.

Either way, those are the two things I spend a lot of time. I got it.

Sriram Viswanathan: Okay. And what about, uh, your favorite, uh, economic thinker that influences your own views on ai?

Ronnie Chatterji: I have many, but I'll say the one who's on my mind, 'cause I just spent some time with that at Stanford. Is, is Eric Sen. Eric is a fantastic economist at Stanford who had really, I mean, he's been really studying AI since in some ways, uh, the mid eighties if you really go back to what he's been doing.

Uh, but he's really been at every technological transformation, a key thinker in understanding how it's gonna affect business and society. And, uh, Eric is something I, someone I pay a lot of attention to, uh, in terms of what he's thinking and, and what he's building at Stanford. Really amazing person.

Sriram Viswanathan: Great.

All right. So you've been at OpenAI for now for a little bit. Um, what is the most surprising trend that you've noticed, noticed at OpenAI since you started? Or, you know, was there a difference between what you thought OpenAI was going to be? From the reality of it after you join either one?

Ronnie Chatterji: Yeah, I'll, I'll say that on the second, and two versions of this one is, I think, I think I said before, the research orientation is just mind blowing.

I mean, I am a university person. I spend my life after doing my PhD teaching research. I love this stuff. And I thought I was making a real transition, you know? And I come in and I feel like, wow, I feel like I'm a student again. And, and that is. I will tell anyone, um, after you've reached a certain age, that is a true privilege to be a student again.

And I do feel that here. Uh, and I think that was a huge surprise, thinking I was gonna come into something that, um, was gonna feel very different. I think the second thing I'll say is, um, the questions you've asked and the question that many people are asking, when I go out and speak around, how's this gonna change our world?

You know, what's gonna be the impact? Inside open ai, uh, whether a person has a very practical job, you know, trying to, you know, sell to enterprise or design the products and more people are using it, whatever it might be, those folks all have those same questions. And I feel like these questions of like how it's gonna shape our society and our lives, uh, everybody's asking those and I, I feel like guess even more conviction that like a big part of our job is to try to figure out evidence-based answers.

They might take a little longer than people want. They might not always be on trend or be able to be summarized in exactly the right LinkedIn post, but like really finding. Evidence-based answers those questions, I become sort of more convinced that that's really important. 'cause I get those questions not just from the usual suspects, but from everyone at the organization.

Sriram Viswanathan: Interesting. Okay. We'll end with the last question. What keeps you awake about ai? I mean, I, I'm sure you can be awake because you're super excited or super anxious, so, which is it?

Ronnie Chatterji:  We, no, I, I'm excited, so that is keeping me away, but I'll say no, we have to get the transition right. The things you asked about are really right on.

I don't have all the answers. I'm working hard to try to develop them, but I, we have to get this transition when AI becomes more powerful and able to do more, um, in the economy, to make sure that people have the right skills, both to do their jobs better and flourish, but also right to take these new opportunities that might exist.

If we don't get that right, if our institutions aren't strong enough, if we're not together on the same team to make that happen, um, it, it will cau, it, it, it, it will cause negative effects that could be avoided. And I think if I, in my little small way, can help avoid a few of those things. And that, that would be the, the accomplishment of a lifetime.

So that, that's really what keeps me awake at night, making sure I can contribute to solutions.

Sriram Viswanathan: Ronnie, it's a great pleasure to talk to you. I learned so much about, uh oh, thanks. What you do and some of the important issues. Thank you once again for just eliminating all the key aspects of AI regulation and the economic impact and some of the things that we all should be thinking about as an industry.

Hopefully I can have you come back, uh, you know, in a year and tell us how you've done in open ai. You know, maybe there'll be other challenges that you're grappling with, but, um, I'm sure Thank you once again. I'm sure.

Ronnie Chatterji: Thanks for having me, and I'd love to come back and look forward to comments and feedback as I'm, as I'm learning, uh, all these new things.

I really appreciate your time.

Sriram Viswanathan: Thanks, Iwan. Thank you once again, Ronnie. I really appreciate it. Cool.