Transcription
Good evening and welcome to the John F. Kennedy Jr. Forum at the Institute of Politics. My name is Amo Ganjiga and I'm a freshman at the college studying computer science and economics and I'm a member of the JFK Junior Forum Committee.
Before we begin, please note the exit doors which are located both at the park side and the JFK street side of the forum. In the event of an emergency, walk to the exit closest to you and congregate in the JFK park. Please also take a moment now to silence your cell phones and join me in welcoming Harvard College undergraduate Matteo Kagliero.
[applause]
Good evening everyone. It is my pleasure tonight to welcome you all to the John F. Kennedy Junior Forum at the Harvard Institute of Politics. My name is Mateo Kagglero. I'm a junior at the college studying applied math to computer science with economics. I'm a member of the John F. Kennedy Junior Forum Student Committee.
We gather tonight at what is arguably one of the most pivotal moments in human history. Artificial intelligence has evolved from being a futuristic concept from to becoming a fundamental part of our everyday lives, quietly but profoundly reshaping economies, warfare, and the very fabric of human society. For the first time since the dawn of the nuclear era, global stability may hinge not just on military strength or economic weight, but also on competing to secure a competitive advantage on this new technological intelligence. All in all, AI embodies something we have never seen before: a shift in how knowledge is generated, a shift in how decisions are made, and ultimately how humanity sees its own place in the world. What will our society look like 20 years from now? And more importantly, how will artificial intelligence have shaped it?
Now, to comment on these ideas, it is my pleasure to introduce you all to our speakers for the evening. Eric Schmidt served as chairman and CEO of Google between 2001 and 2011. Today, he serves as chair and CEO of Relativity Space. He is a founding partner at Innovation Endeavors and is a leading voice in AI and national security in the US and around the globe. Graham Allison is the Douglas Dillon Professor of Government at the Harvard Kennedy School where he served as founding dean and director of its Belfer Center for Science and International Affairs. He also served as assistant secretary of defense in the first Clinton Administration, receiving the Defense Medal for Distinguished Public Service.
Now, without further ado, please join me in opening up tonight's discussion and welcoming our esteemed guests to the stage.
>> [applause] [applause] >>
So, thank you for a generous introduction. It's an extraordinary pleasure to welcome back to the JFK Junior Forum, Eric Schmidt, our colleague and friend, uh, for a discussion of Henry Kissinger, uh, and AI and the future. Uh, as you can see here, uh, this picture of Eric and Henry, uh, uh, they were best friends, uh, a late-developing friendship, but became a very deep friendship in which Eric was generously the tutor for a 95-year-old Henry Kissinger, who discovered AI hanging around the end of a lecture that he had given, uh, when Demis Hibu was, uh, offering to, or beginning to talk about AI, and Henry decided he needed to learn about this. Uh, he called me up. I told him, "Henry, forget about it. You know that you don't have any background in science and technology." I told him, "You don't know the difference between a chip and a potato chip." He said, "That's true, but Eric has promised to to to teach me." So, we're very pleased to have him here. We had him here last year. This is actually, uh, uh, maybe it will become an annual tradition. Uh, Henry passed two years ago last week. Uh, so he was a hundred years old. Uh, to think of such an amazing life over that century. Uh, he was a person who made a huge difference in America's national security and the world, and made a huge difference in the lives of many, many people, uh, some of whom began as his students, some who became as his tutors, uh, and many others.
So, uh, Eric has been introduced. Uh, I would just remind you a couple of things. I would say, first, that Eric was the chief executive who took Google from a startup to one of the great companies of the world, and that's a pretty amazing thing. Uh, secondly, early on, he had identified AI as an arena of the future, and that Google essentially bought up all the super talent in the world that he could find, including, uh, uh, DeepMind, which was actually the company that then brought to Google, uh, Demis Hasiibu, who got the Nobel Prize last year for his work at Google on protein. Mustafa Suleyman, who's now running, uh, uh, Microsoft's consumer AI, was another part of the acquisition, and many, many, many, many others.
So, the other thing to say about Derek, about Eric, and why it's such a pleasure to have him here, is that if one tries to make any sense of all the claims that are being made about AI, most of the people who talk loudest are talking their own book. So when I listen to Sam Altman, or to folks at Anthropic, or folks, even Mustafa, now at Microsoft, they have to talk their companies.
>> You're holding, holding my book.
>> Well, wait a second. They have to talk their company's, you know, business, their future, and they are chasing what they believe is the biggest pot of gold at the end of the rainbow that there's ever been. So, it's pretty hard to tell what they actually think and what they're saying. Eric has kind of graduated to be the statesman in this domain, having had a big stake in it, but now having stepped back, and I particularly after his work with Henry, basically trying to say things as clearly as he can and as clearly as he can see them. So I take him to be a much more valuable source of clues about what's happening than listening to most of the people who are talking their own books. So let's start with Henry, and then we're going to go to AI, and then we'll go to questions in the audience.
So, uh, Eric, uh, you, uh, spoke beautifully at at Henry's memorial about how much impact Henry had on your own life and even on the, you know, the questions that you were asking, the things that you think that matter. So tell us a little bit about more what Henry meant in this relationship, and also for people that wouldn't have the opportunity to to know him, what, what, how might they capture a little bit of this magic?
>> You know, when I, when I first met, and thank you, Graham, as always, it's great to be here. Um, I, I met Henry when he was in his early 80s, and normally men in their early 80s become a little off. And I was struck by how brilliant he was. And so I thought to myself, you know, he, he served in World War II. He got the Bronze Star, you know, immigrated or escaped from Germany because they were a Jewish family. And, um, came to, fought in the war, came on the GI Bill to Harvard, uh, undergraduate and graduate, and ultimately was a professor here. And I tried to figure out, he must have been really smart when he was in your terrace, not that the building existed at the time. So here's a quote when he was an undergraduate at Harvard: "In the life of every person, there comes a point when he realizes that out of all of the seemingly limitless possibilities of his youth, he has in fact become one actuality. No longer is life a broad plane with forests and mountains and beckoning all around. But it becomes apparent that one's journey across the meadows has indeed followed a regular path, that one can no longer go this way or that, but that the direction is set and the limits are defined." That's what he wrote when he was your age. Um, when he was an undergraduate here, he holds the record for the longest undergraduate thesis in the, uh, in the college. After he submitted his undergraduate thesis on Kant and the meaning of the world, they instituted a new rule, which still applies to you, that your thesis can be no more than 350 pages. FYI, it's >> true. >> So, he was clearly an unusually brilliant polymath. And, uh, my own explanation for Henry, aside from, you know, I really cared deeply about him, was that watching his family, and in particular his father, and he, we talked a lot about this, um, see the destruction of the world around him in Germany as the Nazis took over, and when they fled, and he saw the damage to his father and his father's identity and so forth. Henry, after the war, decided that he would do everything he can could to avoid a future war. Now, you can quibble with or argue very much with the things he did, but you cannot argue that his, that his goal was not what I just said. The, the data supports that, the historians all agree. You can disagree with the specific tactics, but at the end of the day, what he sought is he sought a, a world that did not have a World War II. And think about the issues that he faced in the 50s and 60s. He used to tell me about this small group, and he always thought that the most interesting policy things happen with small groups, and this was a collaboration between MIT and Harvard and Rand, uh, which in the early 1950s de, um, um, invented mutually assured destruction. And I said, "Henry, who was in this group?" He said, "Oh, anybody like [laughter] you know, give me a break, right? All the most famous people." So, uh, he was lucky, uh, to be born at the right time. He was clearly one of the smartest people alive at the time. He was lucky to get out of Germany before his family and he were killed. He was lucky to be part of the GI Bill. Um, and he had all sorts of interesting stories. My favorite was one day I told him, "You have to go to the doctor." You know, he's old, and he goes, "Okay." And I said, "I want you to go to the Mayo Clinic." And I said, "Okay." He's a New Yorker, and the people in Mayo are like, you know, Midwestern, soft-spoken, pleasant, nice, not like the intense New Yorker types. He goes to the Mayo Clinic, comes back, and he said, "I loved it." I said, "Why would you like the Mayo Clinic in the middle of Rochester, Minnesota?" He said, "It reminded me of the war." I said, "What?" And he goes, "Well, he was in a, he didn't speak English when he came to the country. And he worked as a, um, uh, in a, um, shaving brush factory where they made shaving brushes. And he was drafted as everyone expected, and he went to war, and he went with a Wisconsin-based American group. So he indexed his identity, couldn't hardly speak English, and nobody could understand his English anyway. He indexed to that group. So he had all these sort of touching little stories. I have lots of stories like that about how he became an American, and of course, became an American citizen, and, you know, the rest of the history.
>> Well, I would say a fantastic person. I had the good fortune to enroll in a course here >> as a student. >> What was that like? >> 1965, God help us, uh, taught by Henry. Uh, and ever thereafter was part of his entourage. And I think the thing that struck me most about him was this strategic acumen. That is, the ability to take a problem and lift it up to its whole 360 degrees of strategic challenge, and then find the points at which the policy process could be impacted. And, uh, I wrote about, uh, for his 100th birthday, which reached to write a little, you know, something that Eric wrote, something I wrote, something, and I said, really thinking about Henry, uh, he was not about, if you think about, he was often criticized for his realism, or ruthless realism, or realpolitik, but if you look and look and see what he was doing, uh, it wasn't, it wasn't raw realpolitik, uh, just for the advancement of the interests of one country. It was always about the construction of a viable order to prevent catastrophic war. He had lived through the experience of the Holocaust and the most deadly war in history. He foresaw the prospect of a nuclear war that could actually extinguish life on Earth. He was caught up in the Cold War, the fiercest rivalry that we'd seen to that point between the US and the Soviet Union. And all the time he was trying to reach beyond that to some way to survive for both the US and for, for, uh, for fellow human beings. So that's, I think that's how he kind of backed into the AI issue, which for him was another generation of a somewhat analogous problem. Well, he had written his undergraduate thesis on Kant and the meaning of meaning, essentially, and, um, he was very, when he went to hear Demis speak, um, it sort of, he immediately got it, and he immediately said to himself, "What does this mean to being human?" And we are today grappling with the question that he foresaw 20 years ago, when we, when we first started working on this. What does it mean to be human in the age of AI? What does it mean to be a child, to an adult, to be a leader? What does it mean for economics? What does it mean for jobs? You know, all of that. But his core argument was that this is an ethical change, in the sense that it's like the, um, uh, the various major changes that we've had in sort of reasoning, scientific revolution, and so forth. Because we, as humans, have never had a competitor that is not human, but of [clears throat] similar or greater level of intelligence, and it is unpredictable what we humans will do. He used to say that what would happen is in magic, when people don't understand things, they either decide that it's a new religion, or they take up arms. And so he would say, "Well, are we going to take up arms to AI, or we make it a new religion?" And I said, "I hope it's a religion." [laughter] Um, because of course, I benefit from the religion, I guess.
>> So, in the book that was published this time last year, called Genesis, which was the work that Eric and Henry and Greg Bundy was doing up to the point at which he, uh, passed. He, here's a passage that clearly Henry wrote. It says, quote, talking about the US and China: "If each wishes to maximize its unilateral position in the AI space, then the conditions would be such for a contest between rival military forces and intelligence agencies the likes of which humanity has never faced before. Today, in the weeks and months and days leading up to the arrival of the first superintelligence, a security dilemma of existential nature awaits us." So you've thought about this superintelligence as an existential threat beyond what what we can imagine. Say more.
>> Yeah. Well, um, [clears throat] there is something which I call the San Francisco consensus. And the reason I call it the San Francisco consensus is everyone in San Francisco believes this. It may or may not be true, but if you go to San Francisco, trust me, that's what they're doing. Besides the usual San Francisco recreational behaviors, and, um, the, the basic build goes that we are, we're, we've gone through a language revolution, you understand that, you know, language, language, ChatGPT, everybody here knows what that is, agents, which are now on the horizon, and these agents allow you to essentially automate tasks. And the key thing to understand about agents is they can be concatenated, so you can, you know, do this, and this, and this, and this, and this, and then they all get applied together. Why am I talking about agents? Because I've just described workflow, and that's what businesses do, what universities do, it's what governments do, so forth and so on. And then the next one is reasoning. Now, reasoning is a higher-order function in humans, and the, the reasoning revolution is just beginning. Um, as of today, it is technically correct to say that the scaling laws that operate in AI have not slowed down yet. The scaling law basically says that if you put more data and more electricity and more chips, you get this emergent behavior one after the other. And you can see that, for example, with Gemini 3, which just came out, which beat, uh, OpenAI 5, which just came out, which beat, uh, Claude 4.5, which just came out, which beat DeepSeek, because they copied it anyway, and blah, blah, blah, you know. So, so it's >> In case you didn't notice, Gemini 3 is from Google. >> Yes. I'm proud to say that we're back in charge, um, until the next one. It's, it's very, very competitive. And what's happening is that these massive data center buildouts, uh, which are, by the way, one of the key drivers of the American economy, are helping not only lift our economy, but they're also building, building this kind of stuff, and it's a whole new world compared to my whole experience. So the question is, what happens over time? So you have language, agents, and reasoning. Well, isn't that what we do, right? We do stuff, we communicate, and we do actions. So the San Francisco consensus is that at some point, that stuff comes together, and you get what is called technically recursive self-improvement. And recursive self-improvement is when it's learning on its own. This is not true today. Today, when you set up one of these huge data centers, you know what they look like, you have to tell it what to learn. But the belief is that this is coming, and there's lots of evidence that this is coming. The ability for computers to write programs, to generate mathematical conjectures, to discover new facts, looks like it's very, very close. Many people believe that there will be new math, design, new mathematicians, AI mathematicians, delivered in the next year. So we collectively as an industry believe that this is going to happen soon. If you ask the San Francisco people, they'll say two years, which is really soon. If you ask me, I double that to four years, which is really soon, right? So, it's happening. It's happening very quickly. I want, and Henry certainly wanted it, we want it to be built with American values and human values. There is a point, in my view, and we talked a lot about this, where somebody's going to have to raise their hand and say, "We just went too far. There's too much danger here. We don't want to give that agency to the computer. We want humans to be in charge of it." It is not agreed to where that point is, but our book spends a lot of time talking about where that might be. Another example would be that you discovered that the computer has decided on its own to get access to weapons. That that's clearly a, like everyone would agree that's not a good idea, right? It's bad enough that humans have access to weapons. Imagine if the computer has it, and what are the criteria by which, and you can think of many other such examples. Fundamentally about human agency. We also spend a lot of time talking about the effect on children. Um, we're running a mass experiment on human development by deploying these systems that are incredibly addictive, whether it's on iPad or phone or, you know, whatever, for young people who may not be, certainly not adults, may not have their own identity, and they can really be manipulated. So what does it mean to be a child whose best friend is non-human? Aside from maybe becoming like a super nerd. But I don't know, what does it mean? We, we have no data. We don't know what it means to young, young boys and girls, to their development, to their ability to associate. Will they ultimately rebel and say, "I only love people. I hate computers." Like, you know, kids rebel. We, we just don't know.
>> So, can you take a next slide? I'm sorry. I had the clicker, but I left it. Can you, can you guys make the next slide go up here? Thank you. So here, the so-called Kissinger challenge. This is Henry, 1969. So Nixon was elected in 1968, became president January 1969, appointed Henry, who was a professor here at Harvard, as his national security advisor. And Henry writes, as you'll see here, that anyone coming to office in the late 60s could not fail to be awed by the unprecedented dimensions of the challenge of peace. And then down to the bottom line: "There could be no higher duty than to prevent the catastrophe of nuclear war." So that became for him a defining challenge in the Cold War, with nuclear weapons arsenals developing, but "no higher duty than to prevent." So how might this apply currently as we try to think about the US-China and AI? Well, let me say that I think there's no higher duty than to preserve human agency and human freedom, right? The things that we value collectively the most. And I think that's going to be a central challenge for all of you when you graduate. Um, all of you will be facing these issues, and they're complicated, and they're subtle. Um, imagine if the internet had been invented by China, and it didn't have the kind of openness that, uh, the internet has today. Was under the Chinese internet. Um, it looks like China is pursuing a different strategy than what I'm talking about. In my most recent visit to China, the way I operate is I ask engineers technical questions, because they don't lie to you, whereas kind of everyone else, I'm not so sure. And I eventually figured out that what the Chinese are doing is they're really, really focused on applying AI in their businesses. And they're going to outcompete with us. We, you know, we will lose to China for their incredible adoption of AI in every product, because they're just relentless, and they work so hard. It's called 996, 9 in the morning, 9 in the evening, six days a week. Illegal, by the way, in China, illegal in America, right? But nevertheless practiced. Um, they're coming. Um, there do not appear to be as focused on superintelligence and the path that I was describing as the San Francisco consensus. Of course, that could change. So it appears that the two are pursuing different paths. One of the questions for you all, and at the graduate level, even undergraduates, given it's Harvard, is start figuring out what happens when these divergent paths hit roadblocks, right? Because they're both have, example would be in America, we have essentially produced no new electricity because it's very hard to provision electricity. China has infinite electricity because of their incredible investment in renewables and so forth. They built something like 120 gigawatts of new renewable energy in the last five years, at some number like that. >> No, no, no. Way, way, way. Every day in China, Thursday, Saturday, Monday, every day another gigawatt of electricity is added to the grid all year. That's pretty extraordinary. >> All year. Every day. So my point is, and here, and here, by the way, just to give you an example, how big is a nuclear power plant? About 1.5 gigawatts. So that's the scale of the electricity revolution that's going on in China, again, using those numbers. How many of those kinds of plants have we built in America? Zero. Uh, and we're certainly losing the renewables race to China for the reasons that everybody here knows. So they have a lot of power, right? We don't. We have a lot of chips. They don't. Um, it sets up the competition, right? And each will pursue different ways. Um, one of the technical questions, there's something called diffusion, where what you do is you take a very powerful model, we'll use Gemini 3, a top one at the moment, um, and you ask it 10,000 questions, and you take the answers, and then the system can learn from the questions and answers enough to mimic, without the expense of doing the big training, the big model. So again, thinking through the strategies that China will pursue, and the, the strategies that are available for America, um, are is probably pretty important. What's interesting is both countries are relying on the private sector to do this. You would have thought in Henry's period that you would have used the government, >> but in fact, our government cannot move this quickly, the compensation systems, and so forth. It turns out it's probably true for the Chinese as well. I don't know at the, at the security level if this is true, but I've not found any large, weird Manhattan-type projects within China, although there's plenty of people in the private companies that are working hard on national security.
>> So, on the AI topic in general, tell us, uh, just give us a minute or two of what is the upside, the things that excite you the most that are one year, two year, three years, that are in line of sight to?
>> Well, the first question is, why is this madness occurring? It must be a bubble, and it's going to crash. No, it's not a bubble. If, if anything, it's underhyped, because you're fundamentally automating businesses. And the reason people are spending this enormous amount of money is to automate the boring parts, or what they view as the boring parts of their business. So whether it's billing, or accounting, or product design, or delivery, or inventory, or management, or whatever, people are automating those. Um, and there's an awful lot there. Think about medicine. Think about climate change, engineering, new science. It's extraordinary.
>> So, what excites you the most of the ones that you see that you have, that you [clears throat] have a line of sight to, that the rest of us probably are not seeing yet?
>> Sorry, [clears throat] I'm sorry. >> I have a cough. I apologize for that. >> We can all see in our own imagination what we're thinking of, and then we'll see what Eric says. Yes. >> Well, when I started, uh, when I was in high school, I was an early programmer, and I delighted in writing code. When I went to college and graduate school, that's all I ever wanted to do. I didn't, I ignored all these history things and things like this. I was the definition of a nerd at the time. And everything that I did in my 20s, which got me to where I am, has now been completely automated. Every aspect of the programming that I did, every aspect of the design, is now done by computers. [snorts] I recently had it write a whole program for me, and I'm sitting there watching it define the classes and the detail of the interactions and so forth as it's generating. I go, "Holy crap, you know, the end of me." And I think watching, I've been doing programming for 55 years. So to see something start and then end in front of your own life, and you're still alive, is really profound, I might say. Now, computer science is not going away. The computer scientist will be, at least until the computer scientist gets replaced, uh, will be supervising this. But the ability to generate code at the power that these systems can do is revolutionary. It means that each and every one of you has a supercomputer and a super programmer in your pocket. Now, nobody here is a terrorist, right? But using, it's always easier to use negative examples. There's plenty of, I'll use a stereotype, young men living in the basement. Their mothers give them food, and they sit there on the equivalent of crypto for 4chan, you know, paranoia, whatever. Pick your, pick your poison. They all have the ability to use these tools to build incredibly powerful systems, right? Cyber attacks, other things, no, whatever they care about. Um, there was, uh, there's some evidence that I think it's Manion, the fellow who killed the insurance executive, um, was into some of this, and there's some people who were looking at some of his writings. Of course, he's in jail now, but that he was somehow influenced with that. Now, I'm not proving causality here, but it's an example of some of the darkest recesses of humanity. You give those people these kinds of tools. We have to be ready. Now, the industry is well aware of this, and we're working on it. It's very important that defensive systems be capable. The eventual solution to AI, by the way, bad AI is AI fighting good AI, fighting bad AI, and that's how it will resolve itself.
>> Okay. Can we do the next slide, please? I want to ask about the US-China rivalry in AI as you see it. So this, apologies if the slide's not clear enough, but what it suggests is that if we take the series of indices, uh, if you look at the performance gap in January of '24, it was significantly larger than it is today. And then, what do we make of this, and what do we make of the likely future?
>> So the chart is correct, but the people who are influenced by this claim that it's not going to be true for long, because the reasoning revolution requires so many chips and so much of the magic that the San Francisco people have invented, using that as a sort of moniker, that the gap will widen. My own view is that the gap will widen, but for different reasons. I think that the Chinese focus is largely, as I mentioned, on embedding AI in everything: smart toasters, cars, so forth and so on. They're moving much more quickly than are the robots. I think that the vast majority of humanoid robots will be Chinese AI-powered and Chinese AI-manufactured, simply because they know how to drive the cost of things down. Their supply chains are incredible. Their cost management, they work so hard, blah, blah, blah, all that kind of stuff. So my guess is that it's true that the gap will probably get larger, but the real question is, will you, as a consumer, ultimately have a better experience as a consumer with a Chinese product than a US product? And the answer is, from a fit and finish, probably the Chinese product, and that's of concern.
>> So let me drill a little further on this one. So one, there's a half dozen questions about which people are making bets, and you've thought about them quite deeply. So one is, are we going to bet on computer chips or stacks or brains? Another one is, are we going to bet closed or open? >> Another one is, are we going to bet, uh, working hard on AGI or diffusion and applications? So if you go across the spectrum there, uh, if I look at the Chinese piece, uh, certainly the guys at DeepSeek, think 200 people, who just have brains, can have a reasoning machine that costs 1/1000th >> of the cost of OpenAI. And now there's six other little dragons coming along from the same, from the same space. So that one makes me worry. On the closed versus open, if I remember our last conversation, you had pretty much concluded that open is going to be closed, but all of our companies are mostly closed. So what about that? And then thirdly, uh, maybe if there's this AGI breakthrough, all the rest of this other stuff won't matter. But if the, if the diffusion and application is already working in these other arenas. So just take us another step on the on the rivalry.
>> So diffusion refers, I'll work backward. Diffusion refers to essentially learning from one model by, by many pairs, and learning from, as we discussed. My own opinion is that, which is not, I'm not sure, but I think so, that the large companies will ultimately not release their largest models. It'll be too dangerous to do so. That they'll subset them. That's what I would do. No, I assume that they'll come to that decision. I think the most interesting question is open versus closed. For those of you who don't have a background in this, open source, open weights. Uh, open source is what I did for decades. Uh, when you use any form of computer, much of the software that you're using was developed by open source, which meant that the source was published, and people would collectively advance it, and there was a whole movement around this, which I was part of. So I'm relentlessly open source in my view. Um, the major companies are closed source, largely for economic reasons. Um, basically, if you borrow $50 billion from the financial market, they do want a return, and saying to them, "Oh, by the way, we're going to give all the models away, and you won't get your return," probably isn't a very good, um, legal or financial strategy. So it, so the American model seems to have emerged as, as closed. Strangely, the Chinese model is completely open, open weights, open, um, source. Why? I don't know. Um, one possible explanation is that the Chinese government has figured out that they were losing in closed source because they couldn't get the hardware, and open source, because it's free, gives them proliferation. So, one of the consequences of open source, open weights, is the vast majority of humans on the planet will use Chinese models. Why? Because they're free, right? And most countries cannot afford the computing and the data centers and so forth. They're just going to take the Chinese models for free and embed them. Now, is that an issue? Absolutely. Because it comes with Chinese values, Chinese training, Chinese biases, and so forth. We would prefer to have it be American. Um, we will see. There is, there are a couple of open source projects in America that I am a supporter of, but they can't raise the $10 billion that are needed from the public markets to get to where they're going. So, they are gems, but they're not at the scale. I've argued that the US government should help fund them. I've argued that philanthropists should help fund them, but I don't really know.
>> Okay, that's very helpful. So, one last question, and then we're going to take questions from the audience. If you were picking two or three questions that a graduate student or an undergraduate interested in this space, AI and geopolitics, should be thinking about, give us two or three questions that we might have somebody that has an answer to, or part of an answer to, the next time you come.
>> So, so a couple. Well, again, we're dealing with some of the smartest people in the world here. So, the first question is, what does it mean to be human in the age of AI? That is a question that you can write many PhD theses on. So, basically, study history, study philosophy, study how people work, study economics, and then figure out what this new technology is going to do. We always in my industry, because we didn't take those classes, we always ignore those things, right? You guys are capable of answering these questions in ways, and if we get out of line, maybe you can use that to call to call us out. Uh, the second question has to do with this China-US rivalry. Why China versus US? The only two countries that are going to matter. And the reason is, you need an enormous amount of money and an enormous number of people. And as much as I like Europe, Europe is not organized, doesn't have enough people, not have enough money [clears throat] to do it. Um, India is not organized enough yet to do it, although they're trying, and most of the other countries don't have enough money, don't have enough the talent, don't have the right universities, and so forth. And then the third question has to do much more with conflict. In a world where you have terrorists who have access to AI, and you have governments that have access to AI, what does conflict look like? What does a terrorist attack look like against a major power? Obviously, I'm not endorsing that, it's a terrible thing. How do we defend against it? Same thing for China versus US. But what about Russia versus Ukraine? What about Europe versus whatever? Trying to understand how conflict plays out in an algorithmic warfare where the AI is driving everything is a very, very fertile area for research and new ideas, and it's just starting. [clears throat]
>> So, good topics for somebody to work on in the interim. Let's start here. Please introduce yourself and a short question.
>> Hi, my name is Teresa. I'm a second-year student here at the Kennedy School, and I'm in Professor Allison's national security class. Uh, we had a class talk about cybersecurity, and most of the tech that underpins us homeland security is in private hands. So what government's governance model do you think will actually help work to coordinate government and private companies, um, to, during, uh, especially during like an AI type of emergency? Um, and given we talk a lot about like US-China rivalry, and despite the technology and investment rivalry, there's also a big difference in government's model. So how do you think the two governance models will lead, uh, in terms of there's a homeland security cyber crisis?
>> Well, first place, I think under the Trump administration, you're not going to see much regulation of AI. That's pretty clear. Um, in China, it appears that they're allowing the, the companies to do whatever they want. Um, although they have the laws about, uh, various things, they don't seem to be implementing them. So it looks like it's an all-out business conflict. The biggest concern I have is cyber attacks. If you can write code the way that I've seen these things write, and every company in my world is now using programmers and AI programmers together, it's extraordinary. It's happened very quickly. Look at Claude Code, for example, the most recent one, which is by far at the top from Anthropic, and others coming. Gemini, of course, claims it has a competitor, but at the moment, Claude Code is a bit better. It looks like, um, if you can write code, you can also write cyber attacks, because the objective function is easy. Just keep writing code until you break something. And if you have enough hardware and enough energy, you can just keep doing that. So I think there's going to be many, many more cyber attacks. Um, and that doesn't mean from governments. It could also be from terrorists and, and bad organizations. And I think getting organized for that would be my primary concern.
>> Peace. Hi, thank you for coming. I'm David Weidman. I'm MPP2 here. Um, you, you mentioned, uh, the strength of current closed-source models. Uh, you can tell me if I'm wrong, but I think open-source models are about a half-generation behind the current closed-source models. Uh, perhaps it's a little dangerous to characterize open-source versus closed-source as an America versus China issue. Um, uh, and my, my fear is that the Silicon Valley people that you're talking about are seeking regulatory capture, and that's why they want to fearmonger around open-source models. Uh, what would it take if we wanted to establish a moratorium on closed-source models, if that's something that we seek? What would it take in order to establish this moratorium worldwide?
>> I, I think I just don't agree with your question. So I'm sorry. Um, I, I don't see the open-source leadership in America, and I see nothing but open-source leadership in China. I think those are the facts. Um, so I don't think there's regulatory capture on the open-source side. Um, and I think again, under the Trump administration, it's highly unlikely that you're going to see significant regulation of the closed-source companies. But I think the closed-source decision is largely driven by economics, not policy. If you see my point. Literally, you just can't. I mean, think about the cost of these things. We're talking about $10 billion, $20 billion. How would you raise that without it?
>> And I think, Eric, this goes back to, if for making our list of questions, I would add, add to it, how do the financing considerations impact the strategic choices? Because if it, if it just happens to be an accident of the financial, uh, structure, that wouldn't necessarily reflect the national interest, that would just reflect the capital markets as they are.
>> I, I think that it's important to acknowledge that America has the most extraordinary capital market financial system in the world, by far. I think the numbers are 60% of the volume and 90% of of the value dollar and so forth is in US dollars. And so countries that are not dollar-denominated do fear, um, the power of that financial market. And we see it in the ability to raise money. When I was in China, it was quite clear in talking to my friends, they don't have access to the depth of the financial market. They literally cannot get the money. The venture financing that was in China is cut by a factor of five since three years ago. Now, there are many reasons for that, not just our world and not just the US. But without that access to capital, it's very hard to to do these large models with these complicated training. Now, you can imagine everything I'm saying changing if the underlying, um, algorithms change. There are people who are working on new non-transformer models, which are less, uh, economically, uh, expensive. There are many people who've commented on the energy efficiency of human brains versus the cost of these data centers. And trust me, our brains are complicated, but they're not energy, they're not energy expensive. So, so again, there, there, there might be a breakthrough that would change the calculus that we're discussing. Go ahead.
>> Please. Thank you, Dr. Schmidt. Uh, my name is Faton Seammanyaku. I'm an MCMPA student here at Harvard Kennedy School, originally from Albania. In the age of AI, you and Dr. Kissinger describe a world where AI systems begin to interpret reality for us. If human strategic judgment becomes shaped, even subtly, by machine-generated frames, what becomes the anchor of responsibility in global affairs? In other words, when an AI-influenced decision produces real-world consequences, who holds the moral burden? The human who acted, the institution that deployed the system, or the algorithm that shaped the perception? And how should democracies redesign their institutions before this ambiguity becomes a geopolitical vulnerability?
>> A very well-phrased point. So I worry that the future of democracy is uncertain, simply because, I'll use an American example. We believe in free speech, and I'm certainly in favor of absolute free speech in America. I'm not in favor of boosted speech or algorithmic speech. So where is the line, right? If, if I say something that's wrong, and then the algorithm, because my claim is outrageous, decides to spread it everywhere, is that appropriate in a democracy? And you can imagine, and again, I'm not making a partisan thing. I think any, any part of the political system can use this for whatever effect. The ability to generate misinformation that people really believe is so simple now that I hope the answer is better education and critical thinking among human beings. But you could imagine if I were evil, which I hope I'm not, and I, I sat down and I started to flood everyone with my particular unique messages, that I could overwhelm your belief in truth by simply relentless copying and repetition. We know that there's something called essentially anchor bias. If you hear it first, then you judge a. So if I manage to get the message to you first that the building is on fire, which it's not, so I'm not committing a crime. Um, if I manage to get that first, you then anchor from that point. You see the danger of this. It's very real. I don't think, and my, my answer to your question is that every, every democracy will face this, and I think democracies will make different decisions based on cultural values and their understanding of the threat. You won't see a single answer from democracies. And by the way, the way authoritarian countries solve this problem is they just ban it. Right? So, a lot of this stuff is illegal in authoritarian countries because they have the ability to suppress speech. And obviously, we, we're all in favor of free speech.
>> Please. >> Go ahead. Yes, ma'am. >> Yes. Thank you very much for, for coming here this evening. So, uh, I'm called Ellen Crane, and I'm a fellow at the Belfer Center. So I'm wondering firstly, could you comment on the role of humans such as you and me in the very long term with regards to AI? What's our, what does our role become? And also, um, we often frame, you know, because of the value systems, the US-China debate as a competition. But is there, is there value to thinking about that in terms of a collaboration? And maybe more interestingly, you know, you mentioned Europe being maybe somewhat disorganized, but they also have enormous strength and enormous talent, like Mistral in France, and so on. And is there something to do there with, um, collaboration?
>> Um, so, so a couple of comments on Europe. Um, I was the first investor in Mistral. So I'm a big MRO fan. Mal cannot raise the money that their US competitors. They have the same problem, and they're working on solutions to that. Um, with respect to US and China, the, um, I started because of Henry, I, I spent about five years looking at Chinese peer competition. And I thought that it would be possible to get closer to China. And I eventually, through his work and others, discovered that the Chinese were more afraid of competing partnering with us than we were with them. So it takes two to tango, right? And
I think it's highly unlikely. I think you're going to see in your lifetime, um, hopefully understanding that we have to coexist, but it's highly unlikely that the two systems will become best buddies for the obvious reasons.
I think on this, this other question about what does, what are humans useful for, uh, in the very long term, um, it's very, very clear that humans are social animals and want to be around other humans. It's also very clear that there are some things that we need, medical care and so forth, which will be delivered by humans. I think that the, it's a rough way to say this, is most other functions will be doable by computers. Will we allow them? That's a question for your research, right? Is where is that line?
My, my old example was the following. This was back when we were building Waymo at Google. So the thought experiment is, New York City is nothing but Waymo and competitor automobile cars, and the engineers at Google and the other companies have figured out how to completely optimize the traffic so that there, the streets which don't change. You have the absolutely mathematically true highest loading per street ever, and it's seamless. And then you have somebody with an emergency, a pregnant woman, you know, whatever, who has to get to the hospital and needs an exception. Is there an exception button in the car which says, you have to violate all the rules because there's something now? If there is not an exception button in the car, then that's dominance of humans by a computer, and humans will revolt. If I go back to the pickaxe versus religion, humans will fight that. That will be seen as oppression by the government, in this case, in the governor of New York City, whatever metaphor you want, and the computer companies themselves.
If the system, on the other hand, is adaptable to human needs and says, "Ah, we have an actual, you know, medical emergency and not somebody who's high on drugs or, you know, whatever, who's just goofing around, or some kid who's, you know, playing with the buttons." Um, and it reasons and it says, "Oh my god, I've got to do everything I can to get the, the person to the hospital." Then it'll be more accepted. So, a lot of this depends on whether it integrates with our human experience and needs. Does it limit our freedom or does it increase freedom? I've come to the view, I'm old enough now to believe that preserving our freedom, our freedom of thought, of motion, of assembly, of gathering, you know, if all of the things, right, are really, really important. If they impinge on our freedom, then they will be fought, and I will be leading that fight.
>> Please.
>> Namaste, sir. My name is Ilma Rose. I'm from India. You mentioned India in your conversation.
>> I'm lucky to be Professor Ellison's a student and to be learning from him. My question is, how can we bring together US and India? India has got some excellent talented people. How can we bring both our countries together in order, uh, to create a world where democracies flourish, and it's a win-win partnership for both our countries?
>> Um, I strongly agree. I've spent lots of time in India, and the literally because of the IITs and the quality of the people, the um, the depth of talent in India is extraordinary. However, the depth of computing is not. Um, last year we did a calculation, there were only about a thousand GPUs in the entire country for 1 billion people. So, I and others have organized to try to to fix that. Um, so I was alarmed with the most recent trade war, uh, which I think has set, uh, India and the United States back, and that needs to get fixed. I don't understand the trade war, and I just, my position is that's hurting us. India is our natural partner, right? It's a democracy. It's a messy democracy, but we are too. Um, in in Silicon Valley, most of the people I work with are of, um, essentially South Asian origin, with India or some of the other countries. So, at least in Silicon Valley, the Indians are flourishing, and I want the tightest possible integration with them.
>> Thank you.
>> Please. My name is Josh. I'm an MPP2 here at the Kennedy School. Eric, you outlined two national strategies earlier. There was pushing the frontier or pursuing adoption, but ideally, you have both. So, thinking about adoption in the US, what are some of the major barriers to enterprise adoption right now? Does that vary by industry, and what should government do to help address those barriers, if anything?
>> Um, the government usually doesn't do a very good job on that. Um, the industry believes that there is something called a technology overhang, that we, your friendly industry, have produced more tools than you are using. Now, that that is our belief, whether it's true or not, you can discuss. Um, and a lot of it seems to be readiness to adopt the technology. The fact that a lot of the stuff is software, most companies don't have very good software people, I'm sorry to say that. Um, the, the changes internally that are required. Um, I personally believe that this technology adoption thing is just a temporary problem, and that as new CEOs come in and winners emerge, the competitive pressure, which are very high in America, will cause that adoption. Um, you don't see very much adoption in regulated industries because regulation is used as an excuse to not innovate. But at least in the hard innovative industries, I think this will get solved. But the general answer is, I would not have the government do very much because I don't think it would help. Um, what would help is just having every business understand that if they want to make money, which is the capitalist vision, they need to use AI much more profoundly. Remember, you can target your customers, you can serve them, you can understand them, you can talk to them, and so forth, all with AI. Um, and a lot of this, there's a lot of negative issues with this. For example, it may lead to layoffs in companies. There's lots of examples where low-end jobs are being replaced by computers, and that's obviously a job loss. That's a societal problem. But from an adoption perspective, the answer is just time.
>> Please.
>> Hi, my name is Aush. I'm a third-year PhD student in computer science doing AI research. Um, my question is regarding, um, you were mentioning China's sort of intense focus on automating AI and business. I guess in the Valley, it seems like, you know, every day there's a new startup that comes around, like it's the big seed targeting some automation for workflow, and we, we also sort of have the nine to nine to six fad. Um, what would you identify as like a key distinguisher in China's AI, um, automating business that you think the US needs to catch up on, or at least in the, the Silicon Valley startups nowadays?
>> What is your PhD topic?
>> Uh, I work on reasoning for large language models.
>> Perfect. Right. So, what is your time frame for AGI?
>> Uh, I'd give it like six to seven years.
>> Yeah. You see, not San Francisco consensus, East Coast consensus.
>> Good.
>> We will see. You may [clears throat] be right. Um, I think that the, it has to do, I think, with the grandeur of the dreams. When I'm in China, I don't hear the same rhetoric as I do in California. In California, it's two years, the world will change, no one is ready, this is humming so fast, and so forth and so on. It's this sort of rhetoric that drives and it's self-reproducing. It's a belief system. It's like a religion. Now, it always takes longer than than the dreamers say. I don't hear that in China. It's, it's, it's noticeably different. So, for example, with reasoning, you know what DeepSeek did with R3? They did a fantastic job. They invented a new way of doing supervised fine-tuning. I mean, really, really clever stuff. You just don't hear it at a national scale. And by the way, DeepSeek is a national champion for China, right? They literally now are, you know, they're on the list of famous companies. They're getting an enormous amount of money. Uh, last time I met with them, they said, "We've solved our hardware problem," which was code for the government is just going to give them a lot of chips, right? Welcome to a communist country.
>> Okay, so this gent lady here, please.
>> Hi, thank you so much for being here. I'm Sonia. I'm a joint degree student with the MBA and, uh, and the policy school. Um, this might be a slightly more out there question and and open-ended, but you talk a lot about what it means to be human. And I think a lot of that is consciousness. Um, and I'm curious when I think about theories of consciousness, which obviously is very ill-defined, AI is hitting more and more of those. And so it, it, like, if anything, we need a revolution in how we think about consciousness. But I'm curious how you think about consciousness in AI. And if you do think that there's a possibility that AI is or will become conscious, what that looks like, what model welfare looks like in that case.
>> Well, let me ask, let me ask you a simple question. So we have, I'm going to give Graham credit for being conscious, >> and in return he appears to be conscious, and in return Graham is going to give it to me. So for purposes of argument, the computer is on the table. Yes. >> And we ask the computer, are you conscious? And it says yes. And so you, being the smart graduate student here at Harvard, um, come up with a series of questions. It answers every question correctly. How do we know? How could you know that the computer is conscious? How would you understand its internal reasoning state? Now, you could instrument the way its reasoning works, which people are now doing. And what they're doing is they're watching what are called super nodes within the, uh, within the weight structure to watch how it's actually making its decision. So maybe you could discover consciousness by inspection, but that's speculation. So I sat down with a bunch of neuroscientists because I didn't know the answer to this question. I said, "How does it, how does consciousness evolve?" And their theory, which is just a theory, was that consciousness evolves when you have dissimilar systems that are working together and growing, and they develop an awareness of the other, right? And that human consciousness evolved because we needed to understand that we were something, right? That the, the development of the id, the identity, and so forth and so on, was necessary in order to command the system. Now, there's no way to prove that. So the answer is, I don't know, but this is a good qu, another, it's a fourth question to ask this audience to work on. And and there's two questions. The first, how does it work? And the second is, how would you verify it?
>> Thank you.
>> Well, I have the unfortunate, uh, uh, responsibility to say that we've come to the witching hour. We dare.
>> Graham. This is so much fun. I love this.
>> Well, you want to stay for five more minutes?
>> Five more minutes. This is, I love these. We'll stay forever. I'm sorry. I just, I was pro, I promised your, I promised your schedulers that I would get you out of here.
>> So please.
>> Short questions and short answers.
>> Sure.
>> Uh, thanks, Eric. DH here from Harvard Business School, a second-year student there. Also have a background in international relations before my MBA. Um, you argue convincingly about the need for an IA, IA, or International Atomic Energy Agency equivalent for AI. So my question is, how does the State Department, how does the Pentagon, and similar agencies need to configure themselves to prepare for this, the coming era of superintelligence? And my second, slightly risky question is, um, if you are looking for someone to help you this summer or this winter to work on this topic, I'd be very happy to [laughter].
>> I love people with courage. Um, so the IEA question, so there's a group of people who are very close to me who concluded that the only solution to the problems that I'm answering is to create the equivalent of CERN. And the idea was that we would get together, all of us, including China, you know, everybody, and we put it all in here, take all the best minds, and we'd all work on this. And, um, as a result, we would build this huge future because the benefits to humanity of this are so great. Start thinking about every disease eliminated, solutions to real hard problems. The, we have Megan here. Energy issues, you know, she's the expert. All of this kind of stuff. I think it's pretty, you probably say it's probably pretty unlikely that's going to happen. So then the next group says, well, then we need an an IEA, which is essentially a mandatory inspection group for nuclear. And what into that scenario, what happens is you'd have to go, and this group would visit the data centers and the algorithms of every company in every country. Now, that occurred, remember, the Vienna group occurred after 15 years of negotiation, including people like Henry, and after two nuclear bombs were dropped on Hiroshima and Nagasaki, which we can all agree was horrific. We've not had such an example. There are people that I know who say the following, and they don't say this with evil intent. They say that we won't, we will have an event that will force this IEA thing, and we hope that it is a Chernobyl-level event. And I said, what does that mean? And they said, not that many people died. Okay. As opposed to a nuclear attack. So there's a, a strain of thinking in my world that is consistent with that. But I think, uh, first, I think nobody knows, and it's unlikely that the countries will, um, put up with that, uh, until there's a real crisis. I used to say that in, I don't know, 20 years after the climate is being destroyed, and you know, the earth is melting, and sea level rise, there's going to be a m, there's going to be a meeting of all the powers where people will say, well, we really screwed this up. We now have to fix it. So I think there is precedent for a global challenge and a resolution, but it's very messy. It's not done rationally. It's not like you and I agree. There has to be some crisis and some political dynamic, and it can be quite severe.
This is an opportunity for the next generation. Just do this right quick. Do I take the three questions? Just do your short questions. We'll let Eric wrap up, please. Yes.
>> Hi. Uh, thank you for your extra time. Uh, my name is Kevin. I'm a second year at the law school. Uh, my question is about automation and jobs. Um, a couple months ago, Sam Altman went on an interview saying that insinuating that if something could be automated, maybe it wasn't a real job to begin with. I want to get your thoughts on that statement and if you agree with the premise that if something can be automated, should it? Um, and relatedly, if you disagree, what are ways that we as a society can decide what we should and shouldn't automate?
>> Sure. Please fit.
>> Um, good night. My name is Kissia. I'm a middle mid-career MPA and a former Googler. So, I'm very happy to see you. Um, my question is regarding algorithm diplomacy. Like Dr. Kissinger used to say that constructive, um, ambiguity and the human pause is really necessary. But when AI is binary, are we getting to a towards a world where escalation is computationally impossible, and does this push for a new algorithm diplomacy that we could push harder?
>> Uh, thank you. My name is Kanesk. I'm from Harvard Business School. Uh, my question is about ethical superintelligence. Uh, first, do why do you not believe that if America develops superintelligence, other nations will follow and then be able to copy it like China did with DeepSeek? And then second of all, if we're trying to embody it with ethics, then if someone else comes up with an unethical superintelligence, isn't that going to be strictly more free and more capable than a constrained superintelligence model that America tries to develop?
>> Sorry, one word about your question again.
>> Uh, yeah, I wanted to ask what your thoughts were on Sam Altman's quote.
>> Thank you. So, um, one of the things that people at this level forget is that human dignity involves purpose, and an awful lot of jobs provide purpose to an awful lot of people, and the loss of those jobs is a major crisis, not financially, but emotionally, in terms of their meaning. So, in order for us to get through this, we're going to have to address that, right? We're going to have to actually do the right thing, and the right thing will be some combination of better tools and so forth. I'm not as worried about the jobs issue because we're producing fewer humans, which I view as a major crisis. Um, collectively, you all are having fewer children than my generation, which is having fewer children than my parents' generation, and so forth. And we need more humans. Um, partly, I'm a businessman, so you need more customers. But the important point is we need more humans, and if we have less humans, then there's going to be open jobs and no people to fill them. AI can help take people who are not ready for a job and get them.
One word from yours, aside from being a future Googler?
>> No, the algorithm diplomacy.
>> And computational. So, in your premise of your question, you describe the algorithms as binary. And so if you think about Kissinger in 1971, when he came up with strategic ambiguity, do you think that a computer in one or two years will be able to invent strategic diplomacy, ambiguity, given that it's already been done before? I think the answer is yes. So I think that, um, the algorithms are getting so intelligent that as long as it's a concept that's been around in the past, it can probably figure out a way to apply it. So I don't agree with you that it's as binary as your question being. And your question was,
>> Uh, superintelligence.
>> Superintelligence. Um, I, I think it's the same answers, right? Superintelligence in one form or another is going to apply. My own view, which is maybe different from the consensus, is that we're going to develop brilliant AI physicists, brilliant AI biologists, brilliant AI chemists, brilliant AI writers, brilliant AI historians. Um, but that that the concept, and and there'll be systems that can drive them. It's not at all obvious that if you were Einstein in 1902 with that amount of math available to you, that you would have had the brilliance to invent, um, special relativity using the algorithms of today. In the industry, there is a view that that's the next really hard problem. There are various theories about it. One answer is you could just do repetition. You could just keep asking questions, you know, the monkey at the keyboard kind of thing, and eventually you do it. Another way is that you could give the optimization function to be curiosity, and if you just wait long enough, it'll discover special relativity. But that's not how Einstein did it. Einstein sat with his little pen and paper at the age of 18 or 17 with a lamp and figured it out. We're not there yet. My own view is that's going to be a very hard boundary. So, in other words, we're going to get to superhuman behavior, which is not the same thing as superintelligence, which I would say Einstein and Kissinger and so forth really were.
Graham, it's such a privilege to be here with.
>> Such a pleasure to have you here, and we look forward to having you back. Let's say thank you. [applause] Very thank you. [applause] Was.