En los medios

Buenos Aires Herald

Levy Yeyati: ‘AI creates tensions that can lead to outsider populism’

El profesor de la Escuela de Gobierno y director académico del Cepe fue entrevistado por el Buenos Aires Herald sobre su nuevo libro, "Automatizados", que escribió jutno a Darío Judzik, decano ejecutivo de la Escuela de Gobierno UTDT, y sobre el impacto de la IA en la política, la economía y la sociedad.

Por Estefanía Pozzo

Eduardo Levy Yeyati is an Argentine economist, academic, and writer. In his latest book, “Automatizados” (Automated), co-authored with economist Darío Judzik, he plunges into the debate surrounding the impact of AI on the economy, jobs, and politics, as well as people’s daily lives.

Levy Yeyati received the Herald at his home in Buenos Aires. With his cat roaming in the background, he answered questions about the book’s key ideas. This interview has been edited for length and clarity. 

What are the main challenges AI imposes on the world of work?

The first thing we must come to grips with is that, long term, there will be fewer jobs. It’s like climate change: it’s going to happen. You can deny it or argue that it’s happening slower than you thought it would, but it’s coming. However, unlike climate change, where the consequences of not addressing it are death, technological progress is impossible to mitigate. We have to focus on adaptation. On how technological change can lead us to a “leisure utopia” and a better society where we share the benefits of technology and not to unequal stagnation where the wealthy consume and many poor people have nothing.

Several aspects of generative AI technology that have appeared over the last few years pose specific questions. For example, new technologies replace knowledge. We must consider whether to reform education so it adapts to technology or change it entirely. Competing with technology or software is a losing battle from the get-go.

Speaking strictly in terms of employment, we have to consider education as a complementary tool to technology, which I think is unstoppable. 

Instead of thinking of programming as the career of the future, which is something we posited 20 years ago and turned out to be true, we have to focus on the human skills or competencies that will complement or, to a certain degree, compete with technology: empathy, understanding, and creativity.

Cover of Levy Yeyati and Judzik’s book, Automatizados.

If there are fewer jobs, workers will have more free time. That’s what you mean by “leisure utopia.” Can you expand on the concept?

Two things are necessary for “leisure utopia.” First, we need to share the benefits of technology. If there is less employment, we have to think of some other way to distribute resources organically. Either by investing more in public goods and services or by implementing some kind of basic income, because, to quote Martin Luther King, we have to turn all citizens into consumers.

In that same speech, however, King said that when you give someone a paycheck, you also have to give them something to do because, given our Protestant ethic, we have organized ourselves around work. If you remove paid work, you have to teach people how to organize their lives around something else, or at least give them time to learn how to do that. That “something else” will be work, but not the work of the Industrial Revolution. It could be like the work Aristotle had in mind when he wrote about an active life, which can revolve around social participation, creation, and even contemplation. These are tasks that can fulfill you but have nothing to do with a job or things we consider productive and are paid for.

We will have to decouple the idea of paid work from the tasks we perform to fulfill or integrate ourselves, socialize, or spend free time. We must simultaneously address economic distribution as well as cultural or socio-ecological matters. The former is essential because, otherwise, we will have a stagnant and unequal system. The latter will probably take one or two more generations to achieve.

How do you think decision-makers see this kind of discussion?

There are differences between countries and generations. Only last year did central countries start waking up to how imminent this technological change is, which is qualitatively different from past changes. One aspect they are particularly interested in is security. Aside from its impact on our lives and leisure, technology opens the doors to changes that can be good but also potentially dangerous, especially when it comes to spreading fake news and disinformation. Language models spill information into the cloud and need help filtering everything out. This means that anything can end up being recirculated.

But there isn’t any discussion between states, and that’s because politicians, governments, and policymakers tend to be quite old. They have a certain inertia in the way they approach new things.

We are witnessing three significant processes of qualitative change taking place at a slow pace that the world is finding hard to tackle: AI, the climate crisis, and demographic change.

These three factors will completely change the economy and society. The way in which we think about societies. Governments need help getting out of their traditional mindset in order to distinguish cyclical movements from trend changes. These are significant tendencies and critical turns. They do what they can, but for now, the approach is quite superficial.

What impact can artificial intelligence have on southern-northern migration processes?

Migration is clearly connected to the demographic issue, one of the three slow-paced processes we mentioned. We see that developed countries are aging and need youth to grow, but this population has a different skin color and causes political changes. We see this very clearly in Europe and the United States. People have to figure out a way to solve that.

AI could temper this demand because it often replaces workers.

If you look at China’s response to aging, it is the first country to have turned automation into state policy. They’ve replaced Chinese workers — not with immigrants but with robots. Somehow, they promoted automation to replace the lack of workers.

One could speculate that AI might lessen the need to open the door to migrations that are politically contentious. In that case, it would be cooperating with the more nationalist politicians who are trying to close borders. But you could also think that by covering these needs, particularly when it comes to low-skilled jobs, AI might temper the bias that these more nationalist or conservative parties have in favor of border control and steer them towards selective immigration.

It could have some impact. That’s still not the case, but we could summarize the situation  the following way: El Salvador might have to begin worrying about the remittances it receives from the United States if AI starts replacing Salvadorans.

What is the relationship between AI and the political process?

That’s an interesting point because AI has more impact on politics than the other way around.

Many economists argue that the same policies of subsidies, taxes, and regulations that developed countries have implemented for years should guide and influence technological progress.

We analyze this idea in the book. I don’t believe it’s the way forward. Governments currently don’t control technology and have very little influence over it. They claim to want to generate more labor-friendly technologies, but all that does is delay things. And even with this strategy of delaying the inevitable, I think there is little that states can do.

However, AI impacts politics because it enhances something that we are already seeing. Large technological companies are powerful entities that specific countries have difficulty controlling, especially in a multipolar world where there is arbitration between countries.

If you had a single world leader or an organically organized world community, you could impose conditions, something that happened, for example, with the minimum global tax.

But that doesn’t happen in practice with technology. On the contrary, there is a technological war between China and the U.S., and there are technological differences and countries in the middle that arbitrate that. It’s tough for politics to influence technology, and much easier for technology to condition politics.  

The book also references AI’s impact on macroeconomics. With fewer jobs, there might be less consumption. How is this tension resolved? 

The way to manage this through policy is by using the redistribution tools we have at our disposal. Suppose you have a tax system based on corporate and personal income. If it’s progressive and there’s income concentration among the rich, they will pay proportionally more, and tax income will rise. You can give back in some way: either through more free goods and services, or through direct transfers. More liberal countries will opt for a transfer, while countries closer to the welfare state will have better public education and services, or social housing.

In order to do that, it’s essential to know if technology will increase the pie size because redistribution is always better when there is more.

There is ample evidence that some of these technologies save costs but do not increase productivity. The main macroeconomic problem is directing technology toward higher productivity.

Suppose this is left to the market. In that case, generating situations similar to those of the Industrial Revolution is straightforward, albeit with very different state institutions. There is a risk of creating a mass of unemployed and poor people, as well as outsiders, who will initially be outside the economic system and eventually fall out of the social system entirely.

And this causes political problems

It does. It generates outsider populism. If you take it to the extreme, as fiction sometimes does, it can eventually lead to civil war. We shouldn’t forget that we live in a very local equilibrium: property rights and contracts are valid without war. To preserve the social organization that, for better or worse, we have in place, where at least we know what we can and cannot do, you need to control people falling out of the system. Politics has to react before this happens.

The first thing you will see is a drop in demand. Further down the road, you will see the disorganization of society. And that disorganization is unavoidable. It is definitely something we should try to prevent and start taking into account from a political point of view.

Why do you say empathy is a refuge from all of this?

Human beings are original. There are many occupations we could label as “made by humans.” They’re going to be very important in the future, and many occupations, or partial ones, will be grouped in this category. Some tasks will be replaced — we’re already seeing this take place— but others will not and are going to be either complementary to technology or something more artisanal and human. I believe this will be the frontier of human work 20 years from now.

How do you imagine the end of labor?

Well, it can be either a utopia or a dystopia. But I’ll tell you what I like to imagine. I’d like to see a return to humanity. We like to think that work dignifies and humanizes us, but most employment is made up of jobs we hate. Work does not dehumanize you, but it conditions you. If AI, or technology in general, allows you to decouple your activities from paid work, you will have more time to go back to searching for who you are as a human being.