Unless there is a redirection of technology in the field of AI, we will get much greater polarization and equality out of it.
In a recent written interview with Economy Chosun, Daron Acemoğlu, a professor of economics at the Massachusetts Institute of Technology (MIT), shared his insights on the prospects of Artificial Intelligence (AI) and its implications for the labor market.
Acemoğlu is renowned as a ‘star professor’ in economics, being one of just 12 ‘Institute Professors’ among the 1,080 at MIT. Recognized for his groundbreaking work in political economy, his research emphasizes the pivotal role of politics and institutions in economics. In 2005, he was honored with the John Bates Clark Medal, a prestigious award for the top economists under 40. Often dubbed the ‘pre-Nobel Prize in Economics,’ this medal foretells success, as 40 percent of recipients later receive the Nobel Prize in Economics, on average, 22 years later.
Addressing the question about the future of AI and the job market, Acemoğlu expressed concerns about the potential impact of AI development that is skewed towards automation, surveillance, and data collection. He stated that if AI development continues to prioritize automation, surveillance, and data collection, it could not only affect employment stability in the labor market but also have repercussions on workers’ incomes. Therefore, he believes we need to redirect the trajectory of technological advancements for the social good.
According to Acemoğlu, government intervention is crucial to achieving this shift. He advocates for the removal of incentives that encourage job automation and calls for active government involvement in steering AI research in the right direction. His viewpoint emphasizes the need to align technological progress with societal benefits.
Acemoğlu recently co-authored a new book titled “Power and Progress: Our Thousand-Year Struggle Over Technology and Prosperity.” The book, which delves into historical examples, offers insights into strategies for addressing the diffusion of AI. After gaining international attention, the translated version of the book was released in South Korea in June, further contributing to the ongoing discourse on the societal impact of technological advancements.
How do you think the rise of AI will change the human labor market?
“There is little doubt that AI technology, in general, and especially generative AI, will transform the labor market. But we do not know how as of now for two fundamental reasons.
First, we are still at the beginning of the process of understanding the capabilities of these technologies. Some claims about how they can perform complex, creative tasks are most likely hype. It will take time to separate the hype from the reality.
Second and more fundamentally, AI is a very broad technological platform that can be used in many different ways. One path will focus on the automation of human tasks, and along this path, AI could damage wages and employment for many different types of workers. On another path, generative AI has the capability to be complementary to workers and create new tasks for them. If this happens, it can contribute to productivity and efficiency and help individuals transition to more meaningful and high-paying jobs.
My fear, however, is that the current direction of technological development and deployment, dominated by a few large tech companies, is much closer to the first path. This is also the basis of my belief that there is a need for fundamental regulation to push AI in a more socially beneficial direction.”
Which jobs will disappear the most, and what new jobs will emerge?
“This is impossible to know to some extent, but more importantly, as I mentioned already, it will depend on the exact direction of AI. But broadly speaking, we know a few things. Jobs easily performed by algorithms and now by large language models (such as simple writing tasks, data analysis, simple programming, or security IT tasks) will be automated sooner or later. There are many opportunities for creating new tasks in creative occupations, but again, the extent to which this will happen will depend on whether large tech companies will prioritize human-complementary investments, rather than the current priorities of data collection, surveillance, and automation.”
AI will likely replace the jobs of low-wage workers. Will AI polarize the workforce and increase economic inequality?
“I believe so. Some argue that AI will be an equalizing force because it will replace white-collar workers and some high-paid workers such as accountants and financial analysts. Yes, there is this likelihood, but the brunt would be on workers engaged in routine tasks because, despite all the hype, the capabilities of generative AI are not enough to lead to the wholesale replacement of high-skill worker tasks.
Moreover, even if some ad workers, accountants, and financial analysts lose their jobs, they can get jobs in other white-collar occupations. This would put downward pressure on the earnings of even lower-skilled workers.
Bottom line: unless there is a redirection of technology in the field of AI, we will get much greater polarization and equality out of it.”
Should governments step in to regulate the replacement of humans by AI?
“I do not think that the complete replacement of humans by AI is on the horizon. But we are likely to see rapid automation of a range of tasks (including those I mentioned above), putting downward pressure on worker wages.
I don’t think governments can regulate automation directly. But they can intervene in influencing the direction of research. They can also remove artificial inducements that exist for automation. For example, in many countries, including the United States, the tax code subsidizes automation while taxing the hiring of workers. Removing these distortions, whilst encouraging broader research approaches focusing on human complementary paths for development of AI, is imperative for the governments of industrialized nations—unless we want to deepen inequality even further.”
Which countries are better prepared for labor market changes due to AI? How is the U.S. preparing?
“This question has two dimensions. Companies and workers. I believe no country is ready in terms of its companies building sufficient expertise in applying AI in a way that will significantly boost workers’ productivity. US companies are ahead of others in some respects, but it’s mostly for data collection and automation. Overall, a recent paper of mine with several co-authors found that even in the United States, less than 2 percent of companies adopted AI technologies as of 2019. This will surely increase, but it will be a slow process.
Turning to workers, I think it is countries that have an education system that imparts flexibility, social and communication skills, and creativity that will be best prepared to take advantage of the age of AI. Here, the United States is not doing well because of low socioeconomic backgrounds. Children are not receiving high-quality schooling in our country. There is especially a shortage of investment in social skills in schools in poor neighborhoods. Some European countries and Canada may be better prepared regarding the flexibility of their workforces.”
Thanks to AI, isn’t there a possibility that children will receive high-quality education?
“The education sector is one of the most interesting in the age of AI. The industry is putting its efforts into automated testing and ways of introducing more online resources. This is a form of automation that reduces the role of instructors.
But there is an alternative path—showing again the two very alternative directions of AI I have outlined above: this would involve much more individualized teaching, implemented in real-time, dynamically in classrooms, with AI-powered tools in the hands of better-trained teachers. Teachers can identify the specific difficulties individual students have and, based on the recommendations of appropriately developed algorithms, decide how to change the curriculum to serve individual students or appropriately form small groups of learners.
This type of individualized education was not feasible to implement on a mass scale before AI. Results from the education science literature indicate that it could have tremendously positive effects on students from low socioeconomic backgrounds. But once again, this is not the direction the industry is going. This is also not the direction schools are going.”
Why is that?
“There are several reasons for that. The first is a chicken and egg problem: without the technologies, the schools cannot do anything. Without the demand from the schools, the industry may not get the proper impetus to develop these technologies. Fundamentally, this path would require not just better AI tools but many more and better-trained teachers (to do the individualized education for smaller groups of students, etc.). This is not a path that is likely when schools are trying to reduce costs, and also it is not likely unless the tech industry wakes up and decides to invest in human-complementary technologies.”
