Interview with Daron Acemoglu: Taking advantage of technology for workers

David A. Price appears as an interlocutor in an interview: “Daron Acemoglu: on Henry Ford, on making AI worker-friendly, and on how democracy drives economic growth” (Economy Focus, Federal Reserve Bank of Richmond, Q2 2023, pp. 22-26). The preface to the interview provides the following summary: “Today, Acemoglu says ‘cheers for economic growth’, but is also concerned that the decisions made by politicians and companies are diverting the profits from this growth away from workers. And, in his opinion, the powerful artificial intelligence technologies that have emerged over the past few years and are embedded in products such as ChatGPT should be regulated taking into account the economic interests of workers.” Here are a few of Acemoglu’s comments that caught my attention:

What type of AI do we need? What technologies of the future will be most useful for society, especially for workers? I cannot imagine any technology that would be harmful to workers over a long period of time and at the same time be beneficial to society. And so I believe that right now we are going in the wrong direction in the AI ​​community. We are going in the wrong direction in the tech community because we are not paying attention to what these technologies are doing to jobs, democracy, mental health and all kinds of problems. So we really need to ask, can we redirect these technologies? …

[O]Of course, workers also need to adapt. And I think that workers who have skills or choose to specialize in things that will be done by machines one way or another will not succeed. Therefore, I think that social skills, social communication, teamwork, adaptability and creativity will be rewarded by the labor market. Just as machines complement people, so people should complement machines.

But make no mistake, it’s not just about these skills. Today and I think in the next 10 years, the economy of the United States will need a huge number of carpenters, electricians, plumbers, many people who are doing very valuable, very meaningful work that requires skill and knowledge, a combination of manual and educational work. We are mistaken in thinking that everything will be digital. And it would be very beneficial for us if we tried to make new machines, including AI, in such a way that they complement electricians, plumbers, carpenters. I think complementarity is really important. …

If you want to think about adding value to employees, you must think about what new tasks they can take on. And the great thing about electric cars – and the Ford plant in the early 20th century is a great example of this – is that they spawned a whole host of new problems.

With the advent of electric machines, manufacturing has become more complex. So you needed workers to maintain the equipment, and then you needed a lot of support jobs: maintenance, design, repair, and a whole host of engineering tasks, as well as a lot of other white collar jobs. So what was really beneficial, both in terms of workers and in terms of productivity, was not the fact that these factories were replacing electricity with some other kind of energy. They have completely reorganized work in such a way as to make it more complex and thus create more profitable activities for the workers.

Not everything was rosy. It was hard work. Compared to today, the workers were exhausted. It was very difficult for them to keep up with the times. It was still much noisier than the factories we will see later. Yes, and Henry Ford himself, especially later in his career, was zealous in anti-union activities. So it cannot be said that Ford was a visionary in every respect. But Ford demonstrated a new type of industrialization that created new challenges and therefore new opportunities for workers.

I am perhaps less optimistic than Acemoglu about the ability of economists and sociologists to predict the current direction and impact of new technologies and suggest ways to redirect these technologies. Even if such an analysis can be made convincingly enough, I am completely skeptical about the ability of the political system to carry out such a policy. Moreover, while the US and perhaps a few other countries are debating what technology could become, other countries in the world will not wait for the results of this contemplative process, but will move forward at the forefront of these technologies.

However, it is interesting to consider what kinds of technology are being encouraged by the current economic and institutional arrangements. Technology is often chasing market size. Thus, investments in health technologies that may be desirable for consumers in high-income countries will generally be greater than those that could save lives in low-income countries. In addition, a healthcare technology that targets a new market of health-insurance consumers may be a more attractive investment than a technology that, say, cuts existing costs by 10%. Similarly, investments in agricultural technologies that affect crops and farmers in high-income countries are likely to be larger than investments that could benefit crops and farmers in low-income countries. As Acemoglu suggests, business leaders in high-income countries may be more likely to prioritize technologies that can replace workers over technologies that empower workers. VCs may be more likely to support digital companies that can start with relatively few employees, rather than supporting companies in industries that require building factories and hiring more workers. A common criticism is that the government tends to favor research projects that are highly likely to yield positive results, and thus tends to emphasize research that offers predictable but modest benefits over research that offers unpredictable, but sometimes much better results. There is much good to think about whether the underlying incentives are built into the existing technology investments in the ecosystem.

In 2019 Acemoglu and Pascual Restrepo wrote “Automation and New Challenges: How Technology Displaces and Restores Labor”. in the spring issue of the newspaper Journal of Economic Perspectives. Interested readers can refer there for more information. From the abstract to this article:

We provide a framework for understanding the impact of automation and other types of technological change on labor demand and use it to interpret changes in US employment in the recent past. At the center of our scheme is the distribution of tasks between capital and labor—the content of tasks in production. Automation, which allows capital to replace labor in the tasks it was previously engaged in, shifts the content of the task of production against labor due to the substitution effect. As a result, automation always reduces the share of labor in value added and can reduce the demand for labor even if it increases productivity. The effects of automation are counterbalanced by the creation of new tasks in which labor has a comparative advantage. The introduction of new tasks changes the content of production tasks in favor of labor due to the recovery effect and always increases the share of labor and the demand for labor. We show how the role of changes in the content of production tasks—due to automation and new tasks—can be inferred from industry-level data. Our empirical decomposition shows that slower employment growth over the past three decades is due to a faster substitution effect, especially in manufacturing, a weaker recovery effect, and slower productivity growth than in previous decades.