Good afternoon, everyone. Can you hear me? Great, great. Welcome to today's Council on Foreign Relations meeting with Daron Acemoglu, institute professor, Massachusetts Institute of Technology, author of the new book What Happened to Liberal Democracy? For visual effect, I'm just going to place it here.
And, of course, 2024 Nobel laureate in economic sciences. Welcome, Daron.
Thank you, Sewell. Great being here.
Thanks. This meeting is part of the C. Peter McColough Series on International Economics. I am Sewell Chan, senior fellow at the University of Southern California Annenberg Center on Communication, Leadership, and Policy. And I will be presiding over today's discussion. Daron, it is such a pleasure to meet you. You have become, in a very short space of time, the economist that perhaps the AI industry most loves to hate. (Laughter.) The Atlantic in 2024 called you the most controversial Nobel laureate in some time.
I don't remember that. But, OK, fine.
And the Economist ran a rather extraordinary screed describing you as overrated, which we will—(laughter)—which we will get to in a moment. So I'm expecting a very free-ranging conversation, but let's start with the basics. I, you know, read your book over the weekend. And I was really struck that this book, Daron, your fourth if I'm not mistaken, is a book in many ways about political theory and institutions. Why did you, an economist, someone steeped in quantitative and methodological matters, why did you see the need to write a political—a book of political theory?
Well, I need to step back to explain why I wrote this book. I actually wrote it in much greater haste, with greater urgency, than anything else I have written. Because I think liberal democracy, one of humanity's greatest inventions, so to speak, is in crisis. And that crisis is deep and is a big risk for our future. Because I think a world in which liberal democratic institutions either collapse or become toothless would be fundamentally different from things that we are used to, both in terms of our freedoms, our ability to participate in collective decisions and public affairs, but also in terms of the economic foundations of how our society works. And I don't think the crisis is something you can just paper over. It requires a serious revitalization of the ideas of liberalism and institutions of liberal democracy. And a lot of that has to go back to questioning where we have partly gone wrong.
And that's why it's very important for us to go back and ask, which are sound foundations for liberal democracy? And what is it that's actually missing in our current liberal democratic discourse and practice? And I think—my belief is that we need a more participatory liberal democracy, more bottom up. And part of my political theory, or political philosophy discussion is to provide justification and foundations for that. So how can we build a liberal democracy that's both aspiring and inspiring, and bottom up, so that it's not driven by some sort of cadre of leaders or activists who say, this is how liberalism should look like. It's more participatory that we can find the means for reaching compromises within the framework created by liberal democracy.
You put—anyone reading John Rawls's A Theory of Justice, you know, in the late 1970s and on, in the twentieth century, you know, might be surprised at some of your reformulations. You don't reject Rawls, but you do, you know, add on things that you think are essential to liberal thought now. For example, an emphasis on the community, humanitarian. Michael Sandel wrote a very praiseful review of your book. You also write about the essential role of economic growth in sustaining liberal democracy. How did you come at these ideas? And why ultimately did you choose to defend liberalism, as opposed to join the very popular attacks on it?
Because I think that the core of liberalism—and, of course, one has to define what that core is, but the core, as I've defined it, is protection of individual freedoms of thought, speech, association, religion, and a commitment to improving the human lot, both for people who are benefiting from economic growth and those who are left behind, I think are foundational for our future. And those are also, I think, the more durable ideas of liberal theory, even though, you know, when you go back to history, you'll see many thinkers deviate from them in some ways. So that's why you need to have a discussion of what you mean by liberalism, or which parts of liberalism you want to sort of focus on. I think those are really important. And if you look at it from a historical perspective, as well as a quantitative perspective, a lot of the achievements that we have made over the last 250 years are related to these liberal ideas.
But it is also very important that we do not just take some already made, prepackaged view of what liberal duties and responsibilities should be. I think that should be a work in progress. And that's the sense in which I actually do disagree with Rawls, even though I respect him more than many other thinkers of that era. I think there are many interesting ideas in him and the aspirations are worthy. But I ultimately don't—sort of, don't like the social contractarian views of liberalism, whereby somehow somebody, via philosophy or something else, tells us, this is the way that liberal democracy should be organized, from a top-down manner. For Rawls, he worries a lot about some aspects of this, but at the end of the day he comes down to something like that with his difference principle, for example. But I think those kinds of things should be much more bottom-up, that different societies can agree different principles for furthering the human lot, within the confines of basic freedoms and this commitment to progress. And that commitment to progress is also the reason why economic growth is so important. I don't think people are going to be happy, I don't think they're going to be satisfied if things were just stationary in their current state.
In an earlier book, Why Nations Fail, you make the case that ultimately in the long run democratic and open societies and systems do better at economic growth. If that's the case—and you've had some critics say it may not be—but if that's the case, why do you think liberalism has had such a crisis of confidence?
OK, that's actually a very, very important point I want to talk about. I see no evidence whatsoever that any one of the ideas or movements that are challenging liberalism and liberal democracy in the West have any possibility of delivering what people want. From Trump to Bolsonaro, from Marine Le Pen to Nigel Farage, all of them are much, much worse. And they are worse because they don't have a workable agenda of creating what I view, and I argue in the book, as the founding pledge of liberal democracy, shared prosperity and voice, people's participation. Worse, their entire institutional nature makes them—and sort of their top-down nature of them—makes them very prone to veer into kleptocracy, corruption, mismanagement. So it's like a triple whammy, not just a double whammy. So there is really no better alternative in the West. We can talk about China, but even if we agree to disagree on what is good, what is bad about China, I think none of the things that the Chinese system is capable of doing can be replicated in the West. So that's not a relevant comparison for the West. So we're really—at the end, it comes down to some form of liberal democracy, either in its old form or in some revitalized form, versus all of the hard left, hard right, entire establishment alternatives to it. And then it is really clear that liberal democracy is where there is a better possibility.
But, on the other hand, despite that—and despite my work, this quantitative work that, in my mind at least, demonstrates quite convincingly that democracy has performed quite well, much better than non-democratic regimes in terms of economic growth, inequality, control of corruption, investment, health, education—liberal democracy has failed, relative to the standards that it set itself. It has not achieved over the last forty years shared prosperity. It has let its ability to provide public services and security to people slip. And it has done really badly in terms of its commitment to voice and self-government.
That's what I wanted to home in on. When you said a moment ago, there is no better alternative, I just couldn't help but think of the times when politicians have said to the electorate, there is no alternative. 2016 Brexit, 2016 Trump, people like Orban and Kaczynski. Right now, the Alternative for Deutschland in Germany, which made huge gains in recent state elections. When you tell voters there's no alternative to, you know, centrist liberal democracy, it seems to me that the—seems to me the voters are saying, hell yeah there, and I'm going to vote for it.
Yes. Well, that's—well, that's because—and I think that's very important. Thanks for calling me out on that. I do not mean that that should be a rallying cry for the voters, I'm using that as part of our intellectual discussion here. And in fact, my call for revitalization is exactly that if we don't revitalize liberal democracy, voters are going to keep on doing that. And when—by revitalizing, I don't mean making better promises. I mean really, structurally, making liberal democracy more responsive to people's needs and getting rid of the big impediments that have emerged within liberal democracy, for both the participation and the flourishing of a significant fraction of the population who now feel left out.
Are you confident that that can happen?
Yes. I am confident that it can happen. I'm not confident that it will happen, but I'm confident that it can happen. (Laughter.)
Look, I think your book will be and should be read by centrist parties across the world. But those parties aren't doing so well right now. The CDU in Germany, traditionally center-right, perhaps the Democratic Party in the U.S., the Labour Party now has a little bit of breath of new fresh air with Prime Minister Burnham. But, you know, centrist parties are hanging on by a thread. So what should they be doing to actually deliver on the promise of better economic outcomes, more affordability?
Well, it's not just better economic outcomes, by the way. That's part of it, but they have to definitely have workable plans to do better in terms of economic outcomes. You can call it affordability, cost of living, et cetera, but it's really what—I think, fundamentally, what people expected from liberal democratic institutions for—have expected for the last 100 years, which is economic growth that becomes widely shared. So that's shared prosperity, wage growth, job growth for workers with different skills. But that's not enough. It also has to inspire. It also has to make sure that people of different skills and different ideological leanings still can feel comfortable within the broad liberal democratic tent. That's what we've lost. And that's why it's both an economic effort, but it's also a cultural, social effort as well.
And at the end of the book, the last chapter I devote to what I think is a workable alternative, which I call working class liberalism. And there are few pieces of it, but the most important one, given your question, is that I argue liberal democracy cannot stand unless it brings the working class, meaning people who still work with their hands in blue collar or white collar jobs and don't necessarily share the same demographic as most people in this room—which is urbanite, well-off, college degree or post-college degree. Those people have to be at the center of center-left politics and at the center of liberal democratic institutions.
And you're actually quite critical in the book of progressive institutions, including some universities, for having put issues of identity above issues of economics and worker issues.
Yes, absolutely. But my criticism is a little different than—it shares some aspects, but it is different than the standard complaints about cultural politics. Because I think what we are calling "cultural politics" is part of a broader whole which is related to the nature of what I call post-industrial society. And post-industrial society has changed both the nature of economic relations, in particular with the rise of digital technologies and globalization complicating the matter of creating shared prosperity, but at the same time it has also elevated managerial technical skills that college graduates have, while providing opportunities for college graduates to become much more influential in politics, much more segregated in the way that they live, in the way that they work, in the way that they befriend different people. So at a time when the economic challenges of creating a shared agenda have become harder, it has also completely shifted to political balance.
And you can see that, for example, as I argue and document in the book, most clearly in center-left parties—the Democratic Party in the United States, Labour Party, which you mentioned, in Europe, social democratic parties in Nordic countries and in Germany, or socialist countries like Spain or France. All of these countries got the majority of their votes from the working class, broadly construed as I'm defining. Now, they get the majority of their votes from college graduates that used to previously vote for center-right. But the college graduates have become much more ideologically aligned with these central-left parties, or the central-left parties have become much more ideologically aligned with that group, which now also dominates both the activists and the members of parliament or leaders of the party. But at the same time, the working class has largely abandoned the center-left parties. And if it wasn't for the union movement, they would have abandoned it even more powerfully. And that has created a real tension, because the center-left parties, which still are committed to fairness, equality, et cetera, don't have the people who are the ones who are suffering economically. So that creates a real schizophrenic issue.
And I think cultural politics, as well as other different aspects of political change that are quite different from the first half of the twentieth century or this first two, three decades of the post-war era, are related to this adjustment to post-industrial society. And now, before we can even contemplate that adjustment, now we have to adjust to a post-AI society. So all of that really requires this revitalization, together with new aspirations.
So let's pivot to technology. Your book talks actually about social media and algorithms. Has the algorithmic age made it harder for liberal democracies to sustain themselves?
Absolutely. But I also reject in the book the view that all of our problems are due to social media. The problems I identify and document in the book all start at least two, three decades before social media. So it's too cheap for us to say, oh, it's just social media's fault. That being said, exactly like you said, Sewell, social media really complicates the matter, because if you have a commitment to voice and self-government that has to function via people becoming informed and active, social media complicates both of those. And it's also absolutely essential for democratic politics, at least the way that I'm defining it, that there should be means and efforts for compromise between different groups.
So that's the big difference between, you know, a social contractarian view where, you know, you might say, we as the leaders of this party or that party know what the general will is, and we're going to follow the general will. So I completely reject that. So you have to reach compromises. But again, social media complicates the problem of reaching compromises, both because of the way people interact, because of the way people don't form the experiences that involve local associations, communities, and things like that. So that's why I argue that we have to regulate social media while still fully protecting individual rights to dissent and individual liberty. So in some sense I am not against absolutism and free speech, but I don't think absolutism and free speech require that we don't regulate the algorithms of social media and the distorted realities that social media creates.
Let's turn our attention to AI. You have become known, fairly or not, as someone who's more of a techno-pessimist.
I would say I'm a techno-realist, but that's a different—
Yeah, well, and we'll—I want to hear about that. But, in particular, you've been very critical of the notion that AI will deliver, A, unbridled prosperity, you know, make life a lot easier. But in particular, very, very critical of the notion that AI will yield huge increases in in labor productivity, which I think is one of the core, you know, preconceptions behind the AI boom that we're seeing right now. Tell me how you arrived at your analysis, and how—and why has your analysis been so critical? Larry Summers, among others, the Economist, have kind of treated you as unduly pessimistic.
I mean, I may be unduly pessimistic. I'm not—you know. I think nobody knows. There's so much uncertainty. But here is where my sort of approach comes from, both philosophically and then I'll explain it economically. Philosophically, it comes from the same place that drives me to be so keen on bringing people with different skills, different backgrounds, different leanings, back into the economic and social center of our society. Because I think human skills are very rich. And they're rich in their diversity and in their interaction with the real world, in ways that are very, very difficult for algorithms or for even very sophisticated models to replicate. Hence, I do not think that it's going to be possible to fully automate even many simple occupations with very advanced models. And that's why I believe we are seeing huge advances in some aspects of artificial intelligence models, but we're seeing very slow spread of diffusion of these models.
And when you look at the data—and I have some recent work on this as well—you're also not seeing the productivity gains, either at the macro level or at the level of firms that invest in AI. It's just a very difficult problem. So that doesn't mean that we're not going to have productivity gains. But it means they're going to be slow in coming. They're not going to be spectacular, like doubling of GDP every three years like some AI leaders are forecasting. And, third, it also means that the exact direction in which you develop AI and you apply AI is going to matter a lot. So those are the basic principles of what I'm saying, essentially. One is, look at the data. You'll see there are no productivity effects yet. There are much smaller employment effects than people were forecasting. This is going to be developing slowly. Be more realistic about what is it that we can do in the medium term, like a ten-year, fifteen-year horizon. And allow for the possibility that whatever is coming from a couple of leading AI labs and companies may not be the only direction in which we should think about AI being developed and being applied for the benefit of human society.
Does that imply that the current frontier lab IPO valuations, the infrastructure build out, the datacenter boom, that it all amounts to a bit of a bubble?
Well, let me answer two versions of that question—one you didn't ask, and the other one you asked. The one that you didn't ask, it certainly amounts to saying I completely reject the notion that a bunch of people, like, probably no more than twenty, know exactly how AI should be developed, with nobody else questioning where AI is going and whether this is the best direction. I cannot believe that we have lived with this kind of arrogance. I think AI is an amazing technology, but is this relentless pursuit of AGI, or actually, I would say, a distorted version of AGI, being the best thing, that's just—like, perhaps they're right. But that's a debate we should have. That's a much more informed debate we should have. That's number one.
That's quite distinct from your question, which is even if they are right in terms of direction of AGI, whether those companies are viable. There, I have my doubts. I wouldn't say they are completely a bubble, there are advances here, but the amount of revenue they need to generate for their current investments to be profitable is just staggering. So a recent paper in Brookings, for example, estimates—and, I think, on the basis of sound and somewhat conservative assumptions—that given the current investments and the projected investment in the next three years, four years, by 2032 the AI frontier labs and leading companies need to generate revenues of the order of about $4 trillion, between $3.7 and $4 trillion. That means 10 percent of GDP. So right now, they are less than one-twentieth of this number, or around that. So that means a growth rate of about doubling every year. It's just, like, a phenomenal, phenomenal growth rate. And if you believe that there are going to be limits to the diffusion, if there are going to be certain bottlenecks, or if open-weight models are going to seriously restrict their ability to monetize the advances, whatever they are, you're not going to think this is going to be possible. And if you think this is going to be possible—which, again, I'm not ruling out completely—but if you think it's going to be possible, you should be really worried about what kind of inequality that implies, 10 percent of GDP going to AI companies.
Regardless of which trajectory that ends up taking, what do you think should be the global—should there be a global governance framework for AI?
Absolutely. So, you know, we're talking about here about American workers and American voters having a say on AI. OK, fine. Imagine we did that. There are six billion people who are not in China, who are not in the U.S., or are not in Europe. What about them? Their lives are going to be shaped by AI. Should we not ask their opinions? Or even Europeans, perhaps we should ask their opinions too. So I think, yes, we do need some kind of global governance.
At this time, I would like to invite members, both in person and on Zoom, to join our conversation with their questions. A reminder that this meeting is on the record. Yes, please, in the back.
Thank you so much for this discussion. My name is Amy Larsen. I'm currently guest lecturing at Berkeley Law and Berkeley Engineering School on AI and geopolitics, having worked in law and politics and tech previously. I wanted to follow up on your charge that one of the things that should be happening more is that democracy has kind of lost its ability to inspire. I wonder what suggestions you might give in order for this to, kind of, regain its footing. And also if you could shape the deployment of AI in a way that helped facilitate this goal, how would you do that? Thank you.
Thank you very much for that question. And I think the two parts of your question are potentially related. The best way to inspire people is to have objectives based on shared moral values, objectives that are feasible, and the right way of bringing people around these objectives. Part of the reason why liberal democracy has alienated people and has failed to inspire is because we've lost that. We have set out either aspirations or objectives that don't appeal to more than 50 percent of the population, or we have promised things that were unrealistic and have turned out to be completely groundless, such as freely trade and everybody will benefit, trust the financial system or trust digital technologies and everybody's going to do fine. So that's why realism is important. That's why bringing people together is important.
And I believe that all of these things right now meet in the context of AI. So we have to set goals for AI around which a sufficient supermajority of people agree to some sort of a consensus, or sufficient majority. I didn't—supermajority. Sufficient majority of people agree. And are consistent, in some inspiring way, about all of us benefiting from AI, and feasible. And that's what I suggest in the book as steering AI in what I call a pro-worker direction. And that's the kind of objective that, if we can convince people it's feasible and achievable, and they can recognize through their own experience or through the mediation of good debate that actually would be much more beneficial for their lives, would be the kind of project that can bring people together. That obviously won't be fun for some of the AI labs.