In short
Daron Acemoglu argues that liberal democracy is in crisis—especially due to declining shared prosperity and weaker democratic participation—and that AI, as currently designed, could accelerate inequality, create a two-tier society, and undermine American democracy.
Guest
Daron Acemoglu, MIT economist and author of What Happened to Liberal Democracy. He studies political economy, inequality, and how institutions affect economic outcomes.
Key claims
Liberal democracy works best when it protects individual liberties, enables “self-government/voice,” and supports shared prosperity. The post-industrial economy has broken the old wage-growth pathway by enabling automation and offshoring. Current AI is a continuation of automation incentives, not “pro-worker AI,” so it may make many workers dispensable. AI will also deepen information centralization (company control) and social fragmentation, threatening democracy.
Notable examples
Taiwan’s digital minister Audrey Tang using tech to bolster participation; early internet/social-media “democratizing” hopes; renewable-energy policy shifts as a historical analogy; autonomous weapons/drones in Ukraine as an AI risk; green/renewables regulation changing technology trajectories.
Written by AI. May contain mistakes. Listen to the episode to check what was said.
Chapters
Tap a time to open that second in VOUnderstanding Liberal Democracy
0:19 to 0:36
Acemoglu explains what liberal democracy should ideally represent.
“It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50-page restoration block, or finally break down that long article you've had open for weeks.”
Understanding Liberal Democracy
1:26 to 2:25
Acemoglu explains what liberal democracy should ideally represent.
“And the point is not to go back to the past or repeat the past's mistakes, but learn from the past successes for building something else, something new.”
Evaluating the Crisis of Democracy
2:25 to 4:17
An analysis of the current state of democracy and its challenges.
“or the best liberal ideas, has been essential for achieving some of these outcomes for a number of decades before plunging itself into crisis.”
The Impact of AI on Shared Prosperity
4:17 to 6:04
Discussing how AI may affect economic inequality and democracy.
“And then the second is, you know, what's going on with the democratic practice?”
AI's Role in Future Economic Changes
6:04 to 8:01
Exploring the implications of AI on job markets and inequality.
“In a nutshell, I say it's the rise of the post-industrial economy and then associated with that post-industrial society.”
Critiques of AI Optimism
8:01 to 10:09
Examining counterarguments against AI's positive economic impact.
“story so centrally is because i see ai especially as currently construed as a powerful continuation of that automation trend.”
AI as a Double-Edged Sword
10:09 to 14:00
Discussing the societal impacts of AI and its potential for democracy.
“Push your limits, train with precision, see the results.”
AI's Role in Democratic Participation
14:00 to 14:58
Exploration of how AI can enhance democratic participation and its misuse.
“AI can be used for bolstering democratic participation.”
The Centralization of Technology
14:58 to 17:49
Discussion about how AI is being developed as a centralizing force, impacting information control.
“I was going to ask you if you could say a bit more about AI as an authoritarian technology, but it sounds like the crux is that it's used to consolidate and control information.”
Reassessing AI Regulation
17:49 to 19:58
Critique of current regulatory frameworks and the need for proactive governance in AI.
“But regulations as a way of steering AI is not going to work.”
Show all 29 chapters
The Future of AI and Public Opinion
19:58 to 23:27
Debate on the influence of public opinion on AI development and historical parallels with renewable energy.
“But regulation is key, of course, if you take regulation to be some sort of government structure innovation around innovation, around digital innovation.”
Navigating Automation and AI Directions
23:27 to 26:57
Exploration of diverse AI directions and the balance between innovation and regulation.
“I love to engage in speculation about the future of AI, even though I'm very...”
Navigating Automation and AI Directions
27:01 to 27:24
Exploration of diverse AI directions and the balance between innovation and regulation.
“It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50-page restoration block, or finally break down that long article you've had open for weeks.”
Concerns about Autonomous Weapons
27:24 to 28:00
Discussion on the threats posed by AI in military applications and the need for regulations.
“You mentioned the development of bioweapons.”
Concerns Over Autonomous Weapons and AI
28:00 to 29:15
Exploring the challenges posed by military technologies in democratic societies, particularly autonomous weapons.
“Autonomous weapons is another thing that gives me headaches.”
AI's Impact on Society and Democracy
29:16 to 31:08
Examining the potential for AI to create a two-tier society and its implications for democracy.
“They don't have a very clear contribution to at least how we judge contribution today, which is economic.”
The Future of AI and Its Effects
31:09 to 32:53
Discussing the implications of AI on social dynamics, education, and mental health.
“Democracy requires that we, all of us, can look each other in the eye.”
Urgency of Democratic Action Against AI Threats
32:54 to 36:13
Highlighting the importance of upcoming elections and the need for a pro-democracy agenda in the age of AI.
“First of all, social media has already been very damaging, in my opinion, for mental health, but also for the way that we communicate with each other.”
Diverse Perspectives on AI's Future
36:14 to 39:58
Acknowledging differing viewpoints among economists and political thinkers regarding AI's trajectory and its implications.
“You can support the center-right or the center-left.”
Building a New Liberal Governing Philosophy
39:59 to 42:00
Discussing the need for a new governing philosophy to address current challenges facing liberal democracy.
“But the best way of getting there is first develop the best foundation models, and then we'll use the foundation models for building pro-worker AI.”
Progressive Policies and the Need for Coalition
42:00 to 44:30
Explore the challenges progressives face in forming effective coalitions and delivering on promises.
“Because I think a neat division of the issues between centrists and progressives doesn't capture the challenges we're facing.”
Progressive Policies and the Need for Coalition
45:03 to 45:29
Explore the challenges progressives face in forming effective coalitions and delivering on promises.
“It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50-page restoration block, or finally break down that long article you've had open for weeks.”
The Role of Charisma in Politics
45:29 to 54:40
Discuss the importance of charisma in political figures and its impact on public perception and solutions.
“And you said that Mamdani is a brilliant politician.”
Working Class Liberalism Defined
54:40 to 56:00
Understand the key tenets of working class liberalism and its relevance in today's political landscape.
“But when you come with too much baggage from the left, I think it's easier than for the other side to paint you as an extremist and those coalitions may be harder.”
Working Class Liberalism: Balancing Freedom and Community
56:00 to 58:59
Learn about the principles of working class liberalism and the importance of balancing community values with individual freedoms.
“for our post-industrial economy and the conflicts in the diverse society that we live in.”
Policies for Economic Improvement
59:00 to 1:00:34
Explore the necessary policies that could enhance job growth and wage improvement in the context of modern labor dynamics.
“Well, first of all, I think labor organization and labor voice is critical.”
Policies for Economic Improvement
1:00:35 to 1:00:50
Explore the necessary policies that could enhance job growth and wage improvement in the context of modern labor dynamics.
“At Equinox, that's high performance loving.”
Policies for Economic Improvement
1:00:53 to 1:01:11
Explore the necessary policies that could enhance job growth and wage improvement in the context of modern labor dynamics.
“It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50-page restoration block, or finally break down that long article you've had open for weeks.”
The Importance of Defending Democracy
1:01:30 to 1:04:12
Understand the critical role of democracy and individual rights in modern society amidst challenges to liberal democracy.
“But the fact is that I don't feel that I understand economics enough to really have a really well-defended...”
Transcript
Automatic transcript. May contain errors.0:00Push your limits, train with precision, see the results. At Equinox, that's high-performance loving. Everything you need to lock in and unlock your potential at Equinox. Start today at equinox.com.
0:16This episode is brought to you by Google Chrome. You think you know a browser, but Gemini and Chrome? That's new. It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50-page restoration block, or finally break down that long article you've had open for weeks. Gemini and Chrome is here for it. Ready to make anything online make sense? There's no place like Chrome. Check responses set up required, compatibility and availability varies 18+.
0:50We're here, I mean, we're talking mainly about your recent book that just came out, What Happened to Liberal Democracy. But before we talk about the 2026 elections, which are obviously looming very large in front of us, and the state of democracy in the United States today, I wanted to start out just by asking, what is liberal democracy supposed to be at its best? Yeah, I think it's very important that we talk about it at its best because my efforts here are forward-looking. And the point is not to go back to the past or repeat the past's mistakes, but learn from the past successes for building something else, something new.
1:37And at its best, liberal democracy protects individual liberties and lays the foundations for peaceful coexistence of people where they can take part in public life, which I call self-government or voice, benefit from public services, roads, healthcare, education, etc. and also take part in the economy in a way that bolsters shared prosperity, economic growth from which everybody benefits. I argue why liberal ideas were very important for these aspirations and for their realization, and liberal democracy as the institutionalization of liberal ideas, or the best liberal ideas, has been essential for achieving some of these outcomes for a number of decades before plunging itself into crisis.
2:45I also want to ask just before we get started at the outset, because most of my interviews are with philosophers, with physicists. I do occasionally talk to economists, but I wanted to ask just what are the tools or data you use as an economist to make an objective evaluation of the state of liberal democracy at any given time? There isn't an objective set of measures. You know, when I say that liberal democracy is in crisis, it's based on a collage of evidence. And it's an impression. So you may disagree. You may say, no, no. This is not a crisis. is just growing pains or something. But the aspects of crisis that I emphasize fall into two buckets.
3:42One, I highlight what I believe to be the fundamental pledges of liberal democracy on which it delivered for a while, but hasn't done so over the last half century. Shared prosperity with inequality, increasing real wages for manual workers and lower education workers actually falling for about 35 years. Public services, which have become much worse and people are more dissatisfied with them. And people's ability to participate in politics at the local, national and associational level, which again, a lot of people complain about. That's one. And then the second is, you know, what's going on with the democratic practice?
4:24So in the United States and in many other Western nations, there's a very high degree of polarization. People continuously express their dissatisfaction with the process, with politicians. They say people like them are not represented. And support for democracy is an all-time low since we've started collecting data on this.
4:49And considering where we are right now, do you think that one of these buckets is in a more precarious position of shared prosperity or our faith or utility or success of democratic institutions? Well, shared prosperity, I think, is really, really axiomatic for liberal democracy. And yes, it is very precarious, especially in the United States, to some extent also in the rest of the Western world, but a little less so. But also, at least in my reading and in the reading of many other economists, AI is going to put much, much more strain on this. So that's why that's one part of the foundations or the causes of the current crisis that really should get a lot of emphasis.
5:44Well, the current crisis, I mean, AI is more of a, I guess, a forward looking question. And I know that you said that you're not as concerned with looking into the past, but is there a simple story of explaining what changed economically over the last? In a nutshell, I say it's the rise of the post-industrial economy and then associated with that post-industrial society. The way that democracies achieved shared prosperity across pretty much all of the European, North American and a few Asian countries that fall into that bucket, was in the context of industrial economics. Businesses would innovate, would invest, would expand, but to do so, they would need more labor for production.
6:56In the framework created by democratic institutions and under the watchful eye of labor organizations, the only way they could do that is by paying higher wages. They couldn't shortchange the workers or coerce them. So that created a powerful pathway from business investment and expansion and economic growth to shared prosperity where wages would increase for all kinds of workers because many of these companies require all kinds of workers, white-collar workers, blue-collar workers, managerial workers, technicians, custodial workers, and so on. That pathway has been largely closed off in the post-industrial economy created by digital technologies and globalization because now companies can expand while laying off workers because they can automate work or offshore it.
7:49automation in particular based on digital technologies has been a powerful driver of growth in many sectors but has also been very very closely associated with the increase in inequality and job loss for workers and that's before ai and the reason why ai comes into my story so centrally is because i see ai especially as currently construed as a powerful continuation of that automation trend. How do you respond to critics? So I've had a lot of people on the podcast, Reid Hoffman or Tyler Cowen come to mind, who are extremely bullish on AI and it bolstering the economy and inclusivity. Well, look, I am also not a pessimist when it comes to AI's capabilities.
8:49But I don't see how you can make an airtight case that AI would create more opportunities for all kinds of workers when its current direction is shaped by AGI, artificial general intelligence. If that word means anything as being defined, and there isn't a single definition, but they all converge on more or less the same principle, means models reaching human capabilities at the highest level and then surpassing them. When that happens, especially given the incentives that companies face for cutting costs, adopting AI, et cetera, they're going to get more and more of the tasks to be done by AI.
9:44Now, it's not in the nature of AI that it should be just an automation technology. And that's why in the book, as well as in my research over the last 10 years or more, I propose what I call pro-worker AI, which would not just automate work, but also create new working capabilities for workers. But that requires a different design, different priorities. And that's not what we're doing at the moment. Push your limits, train with precision, see the results. At Equinox, that's high-performance loving. Iconic spaces that inspire. Personal training backed by real data. Unlimited group fitness classes from yoga and Pilates to strength and conditioning.
10:24Elevate your post-performance ritual with saunas, steam rooms, cold plunges, and more. Everything you need to lock in and unlock your potential at Equinox. Start today at equinox.com.
10:40This episode is brought to you by Google Chrome. You think you know a browser, but Gemini and Chrome, that's new. It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50-page restoration block, or finally break down that long article you've had open for weeks. Gemini and Chrome is here for it. Ready to make anything online make sense? There's no place like Chrome. Check responses set up required, compatibility and availability varies 18+.
11:07are there ways in which ai will also negatively or positively for that matter impact the other bucket that you mentioned earlier the sort of democratic inclusivity bucket and participation of course and that also is not something that's in the nature of ai generally just like taking one step back no technology comes with a pre-programmed social impact technology is broadly speaking
11:46means for us to use our collective knowledge for doing useful things our collective knowledge is vast and there are many ways in which we can use them and the details, who controls technology, what it goes after, all of these things are choices we make as a society. All of those are doubly true for AI. The reason I emphasize issues of AGI versus pro-worker AI is precisely because there There is a mistaken belief that sometimes people refer to as inevitabilism. Everything we're doing with AI is inevitable. That's the only path. I'm trying to fight against that. The same applies entirely when we develop AI as a communication technology or a social technology or a knowledge technology.
12:44Let's go back to the late 1990s, early 2000s. and if you look at writings by tech, some tech people, they thought wikis, the internet, and social media would be hugely powerful democratizing technologies. People were talking of citizen journalism, new democracy online, and they started interpreting the Arab Spring as a result of Twitter and Facebook. Now, in hindsight, that might look a little naive, but it wasn't. It wasn't a crazy idea. The agenda of using digital technologies for more decentralization of power, capabilities, information in a reliable way goes back to the 1950s.
13:36And it was actually possible to use digital technologies and the internet that way. We just chose not to. Why not? Because a few companies that had a different way of making money took control of things. AI will repeat the same mistakes, but it could also, just like early digital technologies, could go in different ways. AI can create much more reliable information, can form online communities that don't cut off people from their real social networks and provide new pathways. ways. AI can be used for bolstering democratic participation. We've seen how a very primitive form of that was done in Taiwan, even before the current AI technologies, for example, under the leadership of their digital technology minister, Audrey Tang.
14:25So there are many things we can do with AI. Again, sometimes people say, very mistakenly, in my opinion, that AI is an authoritarian technology. No, AI is not an authoritarian technology. We're just developing it as an authoritarian technology. We're developing as an authoritarian technology in China for obvious reasons, but we're also developing as an authoritarian technology in the US. The authoritarianism doesn't just come or doesn't mainly come from NSA or the American government. It comes from companies that centralize and control information. That's another form of authoritarianism. Okay.
14:58I was going to ask you if you could say a bit more about AI as an authoritarian technology, but it sounds like the crux is that it's used to consolidate and control information. Exactly. I call that sometimes a better word than authoritarian. We just got into the authoritarian path, so that's why I used it. Centralizing technology. So again, going back to the 1950s, you see the division of many people who were early into computers and digital technologies was that this was a decentralizing movement. But in the background, there was always a very opposed, polar opposed interpretation and agenda, which was centralizing.
15:54It's very funny because early hackers and computer enthusiasts and some of the innovators thought that IBM was the enemy in the 60s and the 70s because they thought of IBM as the ultimate bureaucratic centralizing organization that was trying to control digital technologies and information. Little did they know that IBM then would be a tiny company with tiny control over information relative to the ones that we're dealing with today.
16:29Well, I'd like to, I mean, just constraining ourselves maybe to AI and information technology for the moment. It seems like what you would be proposing is a lot of intense regulation of AI companies to develop pro-worker AI appraisers. No, I actually don't agree with that. Oh, okay. So, a couple of things. First of all, I do not believe, and I think my belief is based on both history and recent evidence, that I do not believe that the government or bureaucrats could become the entrepreneurs and decide how technology is used and how technology is developed. That's never worked. That doesn't mean that the government framework incentives and institutions don't have an effect on the direction of technology, but they would have an indirect effect on the organic process of innovation.
17:38So regulations, if you take them to be hard regulations, I don't think are going to be that effective. I mean, you have to put some guardrails on certain things. I'm all in favor of that. But regulations as a way of steering AI is not going to work. Second, our approach to regulation is completely wrong. I would say we have a reactive regulatory framework. we are dealing with the largest corporations humanity has ever seen we do not regulate them at the antitrust level or monopoly which are different than technology regulation we let them do whatever they want they take freedoms like copying data taking people's data or knowledge or entering into markets without uh without meeting the same requirements etc And then after a while, if there is a backlash or if there are problems, the government jumps in and via a slow process introduces some guardrails.
18:40That's completely reactive. It's ineffective. And it's not up to the standards that would be required for creating countervailing powers against such huge corporations. Instead, what I'm suggesting is more like a proactive agenda. Let's start by first asking what kind of society we want to live in. And given the capabilities of AI, as well as we can understand them, what type of AI would be consistent for that society that we're hoping to create? On both of those, the answer would have to be based on some sort of consensus. That's why public opinion and public debate are very important. Obviously, expert knowledge on what are the capabilities on AI.
19:25but, you know, OpenAI, Anthropic, Google, DeepMind employees have a lot of expertise on what, how AI can be done. They don't have any expertise or they don't have the right to tell society, this is how you should live. That's why you need a broader discussion than just the technologies. Then we think about what are the distortions that are not letting us go there and what are the helping frameworks that could be helpful for us getting there. But it's a much lighter regulatory approach. But regulation is key, of course, if you take regulation to be some sort of government structure innovation around innovation, around digital innovation.
20:07I guess something that I don't really understand is, so I'm with you that it seems like regulation is less successful than we would want it to be. But when AI and the direction of AI is controlled by a relatively small amount of people who may not have liberal democracy at heart when they make their decisions, how is it or why would you expect the democratic process or public opinion to sway their actions? Absolutely, 100 % fair question. But it's not just public opinion convincing one or two people. I think once the conversation changes, the talent that goes into the AI field, which is considerable, could have other priorities.
21:18The reason why we are pursuing AGI relentlessly and investing a trillion dollars in it isn't just that some outman Dario Amadei and Elon Musk believe in AGI. It is that a significant fraction of the very talented people on whom these companies depend on also believe in AGI. And they think, A, AGI is exciting. And B, it's actually good for society, or they've convinced themselves that it's good for society. So if you change that, I think you've also changed the game. But obviously, we also need other government policies to sort of facilitate that sort of transition. We have a lot of examples of this from history.
22:09This is exactly what happened in the area of green or renewable technologies. For a long time, the conversation was, if you want to do something good in the energy area, you have to make fossil fuels cheaper because every country depends on it. And if you suggested renewables in the 1980s or 1990s, most people would have said, oh, that has no chance. But then the conversation changed. Some of the distortions that favored fossil fuels, like, for example, the fact that the damages they created wasn't priced, started being a little bit priced in a few European countries. And starting in Germany and then California, a few European countries started introducing regulations and subsidies to renewable energies.
22:58And it created a complete redirection of technology. There was a boom in new technologies supporting renewable technologies. Solar panels got much cheaper because the technology was enabled and the solar panels got much cheaper because large-scale production started in China and elsewhere. And today, we still have a lot to be done, but in electricity generation, renewables are cost-competitive with fossil fuels.
23:27I love to engage in speculation about the future of AI, even though I'm very... We're all engaging in speculation, even the most accomplished experts. Right. Yeah. But I'm especially poorly informed despite being here at Stanford, which is a hotbed of this sort of activity. Perhaps you've drank the Kool-Aid. I don't know. Yeah, maybe. But one of the questions that comes to mind for me is many of my friends here do computer science. And I'm well aware of just how swiftly advancing AI is at producing code and doing work for them. And something that I wonder is we know how well money motivates people.
24:16With AI developing at the rate it is, I imagine that, let's say, somebody with a lot of money that is directing AI research and policy to some extent, like Peter Thiel. He wouldn't be as dependent on talent because he could pay a lesser quantity of people significantly more to have AI do the work that the large amount of people would have done before. And this makes him less dependent on the conversation. That is true. That is absolutely true. but I don't think, you know, what I'm calling for, and again, this is consistent with history and it's consistent with how technology generally evolves.
25:09I'm not calling for stopping automation. That would be a stupid idea and it would be an infeasible idea. we are very lucky that people don't do the kind of backbreaking work that they used to do in the 1930s or 20s or even 50s and 60s we don't want to go back to the age without cranes and and diggers but what i am hoping is that there will not be just one direction of ai but many directions of AI. So if Peter Thiel gathers around him a group of like-minded entrepreneurs and engineers, all the power to him. That's what in a free society he's able to, and he should be able to. I am not definitely against that.
26:02If that effort then leads to damaging technologies like bioweapons, then of course the government has to regulate those. But when it comes to free flow of research direction, I think that's fine. But underserved directions that are socially beneficial, they do need additional kick from the government. Push your limits, train with precision, see the results. At Equinox, that's high performance living. Iconic spaces that inspire. Personal training backed by real data, unlimited group fitness classes from yoga and Pilates to strength and conditioning. Elevate your post-performance ritual with saunas, steam rooms, cold plunges, and more.
26:47Everything you need to lock in and unlock your potential at Equinox. Start today at equinox.com.
26:57This episode is brought to you by Google Chrome. You think you know a browser, but Gemini and Chrome? That's new. It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50-page restoration block, or finally break down that long article you've had open for weeks. Gemini and Chrome is here for it. Ready to make anything online make sense? There's no place like Chrome. Check responses set up required compatibility and availability varies 18+. You mentioned the development of bioweapons. We just saw, I mean, there was a big New York Times news story a few days ago about how a fully autonomous drone in Ukraine killed a number of Ukrainians that was deployed by Russia.
27:40And that's, of course, a very small and limited in scope example of the existential threat to humans that AI can pose. Is this not a big worry of yours that might sort of put a damper on this whole conversation entirely, that we need to really just slow the deployment of AI? Yeah, absolutely. Autonomous weapons is another thing that gives me headaches. And for democratic society, dealing with military technologies has always been very difficult, even Europe.
28:23exempts many military applications from some of its regulations. Yes, I'm obviously worried about that.
28:34I think we do need to develop new norms about weapons of massive destruction, even if not mass destruction. and there are so many issues that I worry about which I think some of them are already present with drones I think killing somebody when you are in close proximity with them and you see their humanity is very different than killing somebody when you think you're playing a video game there are a lot of issues about you know our humanity and weapons and how technology changes that definitely. I'd like to bring this discussion of AI back to the question of democracy and where we are in the present.
29:23Coming into this 2026 election cycle, how do you think, or how do you see AI playing an important role in what people are thinking going into the voting place or when they're on the internet is it a force for good at present or is it already destabilizing our institutions i think it's already destabilizing a few things
Read the full transcript
29:52but my real fear is forward-looking let me start from my real fear and then we'll come back to the current issues. My real fear, this might, you know, sound something that is a little science fiction, but I don't think it is entirely science fiction, is that we are converging, or we could converge, towards a two-tier society in which a large fraction of the population, 60%, 70%, I don't know, become completely dispensable. Their labor isn't essential. They don't have a very clear contribution to at least how we judge contribution today, which is economic. And then some smaller fraction, 5%, 10%, are essential for AI models, for their engineering, for their design, for their application maintenance and business decisions.
30:50The latter group becomes much richer, much more socially influential, much higher status. As a result, they start intermarrying and interacting less with the remaining group. Such a two-tier society, I think, is the most dystopian thing that I can imagine. It would make democracy, of course, impossible. Democracy requires that we, all of us, can look each other in the eye. one person, one vote isn't just a slogan. It is the essence of democracy. We're all equal when it comes to our voice. Of course, some people will be more influential, more charismatic, but we all endowed with one vote for some fundamental reason.
31:34And I think that would all become a mockery in a world that becomes very two-tier. And in case you think that's too far-fetched, Well, we've taken some baby steps towards that in what I call the transition from industrial to post-industrial economy. There is much less intermixing of people in the United States and in Europe by education. People who are college graduates have very different social circles, marry them among themselves. They live in their separate neighborhoods. They work in their separate firms, and they have very different values today already. So imagine now supercharging that with AI.
32:13And that's what I think is so dangerous for democracy. We're not there yet because we're not even seeing huge money being made from AI yet, huge investments and some money. We're not seeing huge productivity effects. The spread of AI as chatbot is very powerful, but agentic AI is still in its infancy. And very critically, AI model capabilities are still not there that they can sideline millions of workers. But on our current path, we could get there. We could get there in five years, we could get there in 10 years, or perhaps 15 years. But I care about democracy in 20 years' time too. But we are seeing other effects from AI.
32:55First of all, social media has already been very damaging, in my opinion, for mental health, but also for the way that we communicate with each other. And AI could deepen that. We're seeing the effects of AI in the education system that are very problematic. A lot of students are relying on AI, but in a way then that doesn't allow them to form a deep understanding of the subject and their performance is getting worse and worse. And I think most importantly, and this is why in my book, community plays a very important role. Everything we do, exchange of information, our formation of our values, our sense of being part of a human society.
33:41They all start from community. And that requires exchange of ideas, information, and interaction with people. We have much less interaction with people due to social media. People today report many, many fewer close friends than they did 20, 25 years ago. And with AI and us turning to chatbots rather than to our social networks for many things, I think that's going to get deeper as well. And given where we are today, do you see us at a critical point where are we nearing a point of no return or is this a place in time where we still have some leeway in terms of picking a direction in which to go?
34:26How crucial are the upcoming elections to the AI trajectory for our economy? Oh, I think the upcoming elections are very important. I don't think the Trump administration is going to have any meaningful agenda for improving our situation. But I also don't think that there is a threshold beyond which we should throw in the towel. The sooner we act, the better. And there are two reasons for that. Both of them are very major. One is that AI models are getting better and better. And right now, they're not really going in the direction of being more pro-human, pro-worker, pro-democracy. But second, our democracy is getting weaker and weaker.
35:18That's why I wrote this book. I'm kept up at night, sometimes by the failure of AI, by a future shaped by AI, but often also because of the state of our liberal democracy. But liberal democracy is not just important as a general value for us. It's also our only way of reining in the dangerous directions of AI. So the weaker democracy gets, the harder that becomes. So that's why we have to act sooner. That's why also midterms are very important. And broadly speaking to you, for you, voting left is voting more in favor of democracy and voting toward the right is more in favor of authoritarianism and less control over the future of AI?
36:10Well, in general, no. I think in general, people should be able to...
36:19The liberal tent is a big one. You can support the center-right or the center-left. In Europe, for a long time, Christian democratic parties were very central to building a welfare state and supporting shared prosperity and individual freedoms. But right now in the United States, because of Trump's takeover of the Republican Party, the Republican Party has become an entire establishment and more authoritarian. And therefore, at this juncture, yes, I think it is really the hope is that Democrats are going to rein in some of that. And I think also the agenda of passing better laws for creating a framework for AI would benefit from a Democrat government or the party from the Democratic government.
37:13I'm using the word democratic in two different senses. I'm trying to avoid that. That's not to say that there aren't Republican lawmakers who are also very much worried about the same issues. So I've testified a number of times and there are many senators from the Republicans who have concerns that are valid and they would actually be part of a coalition to perhaps write better laws, but not at the moment, not with Trump in power in the White House. This is the last question that I'll ask about AI because there are plenty of other topics for us to cover. But are there any other economists or political thinkers whose work you really respect who have a very different opinion on the trajectory of AI from you and believe that even following the current trajectory, they have an optimistic outlook?
38:18Yes, absolutely. Okay. So one thing I emphasize all the time, and sorry, I didn't do it as much this time, because we started from my book on liberal democracy. But whenever there is a conversation about AI, I say, there's so much we don't know. and you really have to recognize that uncertainty. So here are three ways in which what I've said could be wrong. One is that what I'm calling pro-worker AI is impossible. And whatever I say, whatever policy we do, we're going to go to AGI that will replace humans. That's a possibility. I can't rule that out. I don't believe it's the modal scenario, but it's a possibility.
39:01Second, it is also possible that the question that you asked at the beginning, that the current direction somehow will create more jobs that I'm allowing. Again, I cannot rule that out. I really think it's unlikely. That one's even more unlikely because if really it's AGI, meaning that models can do everything, including social relations and understanding the theory of the mind as good as humans, that's a very, very important part of the skill. If you don't do that, you cannot really perform many of the occupations that are so central to the economy. If they start doing that, then it's hard to imagine that even new tasks and new things would create big advantages for humans.
39:49But it's possible. A third possibility is more interesting and intriguing, which is that I've also gotten the pushback from some people in tech that, yes, you're right, we do need pro-worker AI. But the best way of getting there is first develop the best foundation models, and then we'll use the foundation models for building pro-worker AI. So in that telling, the current direction isn't cementing the automation imperative. That decision is going to be done later once we reach higher levels of intelligence in the artificial intelligence models. Again, I don't find that as compelling, but it's possible.
40:41I really appreciate how careful you are and that you're considering all the sides of things and not just only looking in one direction. I'm trying. I'm trying. What are the other key policy choices and inflection points that we're facing right now in these elections as far as the future of liberal democracy is concerned, other than AI? My book, in my mind, is an effort to invite the reader and other people to start dreaming of and constructing a new governing philosophy for the center left. Working class liberalism? Working class liberalism is the idea I give to it, but I don't care about labels and I don't care about being right.
41:32if I'm wrong and there's a different governing philosophy that's going to be effective, sign me up for it. But I think we do need that governing philosophy. So a big battle has to be within the hearts of Democrats in the United States or labor in the UK, et cetera. And this debate between sometimes what are called progressives and moderates or progressives and centrists is actually very important and very interesting. And unfortunately, I think it's slightly confusing. Because I think a neat division of the issues between centrists and progressives doesn't capture the challenges we're facing. Progressives are right that we need radical policies, especially on the economic front.
42:25But they are not sufficiently attuned or attentive to forming a big, broad coalition, bringing back the working class back into the fold of the Democratic Party, which is very important. And then the other issue is that it's not enough to emphasize these questions, but you need to have a workable plan such that you can actually deliver on the promises when you come to power. I think, for example, Mamdani captures that dilemma quite well. I think he's a brilliant politician. I'm very impressed by him. And he's talking about some of the very important issues like affordability. On the other hand, if you look at his solutions, I think they're not up to the task.
43:12You're not going to solve the affordability problem, which I think is just a code word for the inequality and the collapse of shared prosperity, living standards and opportunities for people in New York and beyond. You're not going to solve that by having subsidized transport and sit around grocery stores. You need something much more comprehensive. And if you promise to solve these problems with such simple solutions and they don't work, that's going to tarnish the agenda of really trying to do something more radical. And I think that's where we really need to come up with the right philosophy that can bring people around.
43:49And I think the most important thing I would say there is that liberalism, at least as I see, never has the answers. You shouldn't turn to liberalism, liberal ideas, or whatever philosophy that we like as having the answers to complex questions. If any philosophy claims to do so, I think it's as good as the dustbin. What we should hope for is that our philosophy provides a framework within which we come together around shared values and we will reach some compromise or communication and dialogue. I think that's the way I want working class liberalism to be perceived. Push your limits. Train with precision.
44:31See the results. At Equinox, that's high-performance loving. Iconic spaces that inspire. Personal training backed by real data. Unlimited group fitness classes from yoga and Pilates to strength and conditioning. Elevate your post-performance ritual with saunas, steam rooms, cold plunges, and more. Everything you need to lock in and unlock your potential at Equinox. Start today at equinox.com.
45:00This episode is brought to you by Google Chrome. You think you know a browser, but Gemini and Chrome? That's new. It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50-page restoration block, or finally break down that long article you've had open for weeks. Gemini and Chrome is here for it. Ready to make anything online make sense? There's no place like Chrome. Check responses set up required. Compatibility and availability varies 18+.
45:28just as a parenthetical question i'm curious uh well actually i the other day i interviewed an evolutionary psychologist uh from cambridge and and and then the other day i interviewed an evolutionary biologist from harvard and it was interesting to think about just how you're an economist i'm i'm studying philosophy here at stanford but what's more interesting to us than anything else is social dynamics and people. And you said that Mamdani is a brilliant politician. And I'm just curious why you feel that way about him. Not that I disagree, but I'm just curious. Because he's got charisma, he has zeroed in on important problems and he can communicate those problems and empathize with people.
46:21I think those are very, very important qualities. Now, I'm not saying that a politician who doesn't have those qualities cannot be successful. But those are very helpful, especially at a time when we need to rebuild community. And I'm really glad you're saying evolutionary biology, evolutionary psychology, and philosophy in one breath, because those are some of my inspirations in this book. I draw a lot on philosophical ideas. I disagree with some of these existing philosophical lines. But I also get a lot of inspiration from evolutionary biology, evolutionary anthropology about the role of communication and ID sharing and social learning in society, which are very important parts of the foundations I offer for liberal ideas success.
47:13I'm glad that you appreciated that. But also, I find it a very interesting distinction to draw between being a good politician and someone who can offer practical solutions. And that Mamdani is a great politician because of his charisma and how he connects with and inspires people and builds community. But that doesn't necessarily mean that he's capable one way or the other of But you have to be. I think, look, I tell you somebody else who is a great communicator, has great empathy and some amount of charisma. Nigel Farage. Go and watch him. And you'll see he's very good at talking to many ordinary folk.
47:59he comes across as sincere and he zeroes in on some problems. And obviously, the solutions he offers to the problems in the UK are horrible. They'll make the problems much worse. And he's responsible partly for the Brexit fiasco. So just having that charisma and being a good politician in terms of appealing to people isn't enough. Trump, I still don't understand why he has charisma, but obviously he does. And he's been such a destructive force in US politics. When you say that you don't understand why Trump has charisma, do you mean that he doesn't appeal to you or you don't understand how he can appeal to anyone, but you accept that he does?
48:45No, no, he appeals. I understand why he appeals to people and I understand he's talking to people's grievances, but I never thought that he was charismatic. as opposed to Nigel Farage who I see as charismatic in his own way or Mamdani who is definitely charismatic right he smiles it makes you want to smile exactly no no Trump doesn't smile exactly Nigel Farage smiles a lot but perhaps I like people smiling so perhaps I'm just too much influenced by that and it goes back maybe to some evolutionary biology and psychology that when Mamdani smiles we want to smile That's one reason we like him. Now, you characterize this inflection point as a sort of debate between progressives and centrists on the left, because this is an argument that you said will take place within the hearts of Democrats.
49:42And what that makes me wonder then is why these elections in a couple of months are so important when it seems like actually the primaries in the Democratic Party were where it was very important. You're right. You're right. But primaries are always in the shadow of the election. And that's actually a lesson that Democrats learn and forget, you know, and Republicans too. elect people who are going to be electable. In this primary, you have to nominate people who are going to be electable. And I think that's very important. A lot of the mistakes that I think progressive have made is because they were talking to their base.
50:32Like, for example, if you wanted to have a much more constructive debate on immigration, which would bring people from left and right together. You wouldn't start with open borders. You would start with, you know, sort of how can we make immigration work both for the American people in line with their priorities and also deal with where we need labor, for example, in some high-skill areas and also issues of humanitarianism, authoritarianism like refugees, but you would find a frame that would appeal to the right as well as the left. Same thing with the primaries. If you think the primaries are just about your party, you're going to go wrong.
51:26And each time I've tried to sum up and make a generalization, I've been wrong so far in our conversation. Am I right? I'm probably not to ask whether you think that on balance, the more centrist rather than progressive candidates on the left have the more practical solutions to the problems, even if the politicians farther on the left are very charismatic in certain instances? Yes, I think the left is more charismatic because they're coming up with radical solutions. And I think they've been better at identifying the problems. But the centrists, I think, have a better chance of finding a path to appeal to voters on the basis of some of those promises, except for the most left-leaning places, and also create a governing coalition around it.
52:31So the hope is that there will be a mixture or a synthesis of progressive ideas and some of the centrist positions. But centrist have also tarnished themselves over the years by not acting, not accepting the inequalities and all of the problems. So that's why it really isn't just siding either with the centrists or the progressives. Okay, just to clarify something, what it sounded like was the centrists have a better chance of winning an election against somebody on the right. And that makes them, in a certain sense, much more practical than the progressive candidates. Correct. But are you also saying that the solutions proposed by the centrist candidates, assuming that both were elected, or assuming like looking at two different worlds, one in which the progressive is elected, one in which the centrist is elected, the centrist solutions are much more likely to succeed?
53:35No, not at the moment. I think centrists in the best of worlds, centrists would find a coalition that includes people from the right as well that would support workable solutions. And I think that has to rely on some of the ideas from progressives. Like, for example, minimum wages. I think it's very straightforward. Minimum wages can work. They have worked. They are not going to be a solution to our problems, but they would work much better and would be much more durable if there is a coalition that supports minimum wages, both from the right and the left. And in fact, if you look at people who are, quote unquote, working class, lower education than college and working in manual or office jobs, there is pretty widespread support on minimum wages.
54:33So, you know, that's the kind of coalition we need to make work. Now, if Mamdani can do that, I love it. But when you come with too much baggage from the left, I think it's easier than for the other side to paint you as an extremist and those coalitions may be harder. So that's the one consideration you have to bear in mind. Real politic may be complex. But I think some of those can be bridged. But I think the important issue, and again, this is the historical part of my book, when you go back, successes of liberal democracy are much more likely to be found in places where liberal democratic ideas were supported by a broad coalition, both from the left and the right.
55:30We haven't really spoken that much, but now we have a little bit of time before we finish, about more broadly what working class liberalism is. And as a political philosophy, what are its key tenets and why do you think that they're vital moving forward for a healthy left? So before I describe that, let me say that what I'm trying to do is think of a governing philosophy that is appropriate for our post-industrial economy and the conflicts in the diverse society that we live in. within that context i think working class liberalism prioritizes bringing the working class back into the liberal coalition but does so while remaining faithful to the most important principles of liberalism it starts with shared prosperity so jobs and wage growth and opportunities is critical.
56:36Second, it creates a broad enough tent that the working classes that often have more, you know, lower education communities, that often in the United States have more conservative values can be brought back into that tent. But it remains very committed to the freedom of speech and protecting the individual. And this is the part about finding the right balance between community and freedom. Community is essential for our values, for our existence, for our social networks, for our ability to do self-government, starting at the local level, for our associations, including labor organizations. On the other hand, communities can often, do often restrict freedoms of individuals.
57:25So you have to balance that. And that's where the liberalism part of working class liberalism is key. You have to protect the individual's right to speech and belief and association and also right to exit those communities. So that's why it's a work in progress. I don't think there is a formula, especially not a formula that will apply in the kinds of society that we live in right now that has high levels of polarization, very complex levels of cross-cutting differences. On the other hand, we have managed to do that in many instances, even in the recent past. State-level referendums on gay marriage were an excellent illustration of how you build consensus that was bipartisan in some sense.
58:14the Irish citizen assemblies in the context of abortion, the topic that everybody thought would destroy Irish society, and Ireland came out of it stronger, was another example, was based on dialogue, understanding the other side's perspective, building on shared values. So I think that's the kind of thing that working class liberalism has to appeal, so that the working class is back and part of the coalition supporting liberal ideas and shared prosperity. And as an economist, you seem to be the right person to ask this. Other than setting, again, our discussions about automation and AI aside, what are the policies that liberals endorsing this platform or this philosophy would enact or could enact to improve jobs and wage growth?
59:09Well, first of all, I think labor organization and labor voice is critical. Trade unions may not be the main form for this because the way that they develop was much more around blue color workers. And we need now solidarity among workers in much more diverse situations. Building that solidarity isn't easy. So we need new kinds of labor organizations. Those labor organizations have to be much more attuned to issues of AI, but we also need legislation. In the United States, current legislation, for example, doesn't allow industry-level bargains, and that may be very necessary for new kinds of labor organizations.
59:53Taxation is another big issue. We have a very high level of public debt, and we have huge levels of inequality. and one of the things we do when we look at our tax code is we don't tax capital income. Capital gains are taxed at a much lower rate. A lot of capital income completely remains untaxed. That's both a problem of inequity but it's also inefficient. I explain in the book based on some of the research I've done is how this asymmetric distorted tax code actually encourages excessive automation not just inequality. There's another thing that we need to have policy. And I think, again, there could be bipartisan support for policies like that.
1:00:34Push your limits, train with precision, see the results. At Equinox, that's high performance loving. Everything you need to lock in and unlock your potential at Equinox. Start today at equinox.com.
1:00:50This episode is brought to you by Google Chrome. You think you know a browser, But Gemini and Chrome, that's new. It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50-page restoration block, or finally break down that long article you've had open for weeks. Gemini and Chrome is here for it. Ready to make anything online make sense? There's no place like Chrome. Check responses set up required, compatibility and availability varies 18+.
1:01:17And, well, I'm glad that I asked that just because people often ask me or expect me to voice stronger political opinions on the show or just with my friends or in conversation. But the fact is that I don't feel that I understand economics enough to really have a really well-defended... And of course, ask three economists and you're going to get four answers on some of these questions. Right, exactly. A similar question. What are the policies or directions in which the left should be going to find this or strike this appropriate balance between community versus freedom? I think we have to respect community more.
1:02:04If you go back to the forming stages of liberal ideas, community was very important. And somehow we have sidelined it. And I think community is critical if you want self-government because people have to start exercising their voice and their ability to shape their lives starting at the local level. But this is where a commitment to a core set of individual values becomes very important. You cannot allow communities to take people's right of exit. If you're not happy in that community, if you don't share their values or you think they're restricting you, you should be able to leave. And that is a, I think, founding principle for liberal ideas.
1:02:55Well, with this last minute or so that we have left, there are a lot of elections going on in November. Each one has different issues. But are there any sort of key thoughts that you think voters should have in mind when they go in to place their vote? So yeah, I think this is a very important election. And if I think about it myself, the most important thing for me is defending democracy. And democracy is one of the greatest achievements of humanity. Think of our ancestors and the ways they lived, the ways they dealt with conflict. the hierarchy, the violence, and we've built much more peaceful societies that respect individual rights, individual freedoms, are consensual in many important ways.
1:03:56We talk to each other, even with the crisis of liberal democracy. That's just an amazing achievement. So we have to defend that. Well, Darren, it's been a pleasure speaking with you today. Thanks so much for joining me. Thank you, Robinson. It was wonderful.
1:04:18Hey, I'm running out to drop off a stack of letters and packages. Need anything shipped while I'm gone? Save the gas. Stamps.com lets you print postage and schedule pickups right from your desk. Sync your contacts into the address book, get detailed accounting insights, and get notified the moment your certified mail is delivered. All from my desk? All from your desk. Join the 4 million businesses using Stamps.com right now with code podcast and get your free welcome gift. Stamps.com, code podcast. Taxes and fees apply. I'm not giving up. I am selling the building. The final season of FX is the bear.
1:04:57The restaurant is flooded. Everything's either going to be okay. No. Or not. We are outgunned and we are outmanned. We have each other. FX's The Bear, the final season. All episodes now streaming on Disney+.
From the publisher
Daron Acemoglu is an Institute Professor of Economics at MIT and a recipient of the 2024 Nobel Memorial Prize in Economic Sciences, which he shared with Simon Johnson and James Robinson for research on how institutions shape prosperity. He is also the bestselling co-author of Why Nations Fail, The Narrow Corridor, and Power and Progress. In this episode, Robinson and Daron discuss the future of liberal democracy in an age of artificial intelligence. More particularly, they talk about how AI could undermine democratic institutions, whether tech billionaires have accumulated too much power, and why Daron believes current approaches to AI could produce an increasingly unequal, two-tier American society. They also discuss the importance of the 2026 elections, Zohran Mamdani, disagreements between the center and far left, and Daron’s proposal for a renewed “working-class liberalism.” Daron’s latest book is What Happened to Liberal Democracy?: Remaking a Politics of Shared Prosperity (Dutton, 2026).
Daron’s Website: https://www.daronacemoglu.com
What Happened to Liberal Democracy?: https://a.co/d/0f0qGMxt
OUTLINE
00:00 What Is Liberal Democracy?
09:30 How Does AI Undermine Democracy?
18:30 Are the Tech Billionaires Out of Control?
24:50 AI Is an Existential Threat
27:18 How AI Will Turn the USA into a Two-Tier Dystopia
35:17 Other Views on AI
38:19 Why the 2026 Elections Are So Important for the Left
42:30 What Makes Mamdani a Great Politician?
45:58 The Far Left vs the Center Left: Who Should Win?
52:00 Working Class Liberalism
