Artificial Intelligence and labor markets: navigating an uncertain and fast-moving landscape

By Veronica Amarante PEP Research Fellow, Gender Committee - Professor, Instituto de Economía, Universidad de la República, Uruguay and Guillermo Cruces PEP Research Fellow, Resource Person - Experimental Group, Deputy director, CEDLAS, Universidad Nacional de La Plata, Argentina

The unprecedented developments in artificial intelligence (AI) technology and its recent widespread availability hold transformative potential to reshape global economic dynamics and labor markets, though opinions on its impacts vary greatly. Transformative or not, the irruptionof AI has sparked ample public discussion as well as the rapid development of theoretical and empirical approaches to understand it. 

In this context, PEP performed a review essay that discusses recent perspectives, theories, and empirical evidence on the matter. It also offers a perspective on the implications of this technology and of the policy discussion it has sparked, especially from the perspective of developing countries’ economies. The latter is motivated by the relative scarcity of such perspectives in the current public debate. As a result, we lack informed understanding of the potential impacts of the interaction of labor markets between developing and developed countries. Nor have we studied the varying effects of such interaction within unique country contexts given the different levels of development in each.

This motivation has also led PEP to replicate a survey implemented by Chicago Booth’s Kent A. Clark Center for Global Markets on how AI will change the labor market, but adding a developing country perspective to it. This blog explores some of the results brought forth by Chicago Booth´s and PEP´s surveys.

Key Highlights from Chicago Booth Review´s Survey

The magazine turned to its panel of economic experts for insights on the impact of AI on the labor market. Specifically, they asked 50 European and 41 north American experts, including Daron Acemoglu and David Autor, among others, on the AI impact over the next 10 years on: (1) the earnings potential of high-skilled workers in advanced countries; (2) the levels of uncertainty about the likely returns to investment in education; and (3) income inequality. The following figures portray some of the survey´s results.

Figure 1.Use of artificial intelligence over the next ten years will have a negative impact on the earnings potential of substantial numbers of high-skilled workers in advanced countries.Responses weighted by each panelist’s confidence.

Source: Chicago Booth’s Kent A. Clark Center for Global Markets

Figure 2. Use of artificial intelligence over the next ten years is likely to have a measurable impact in increasing income inequality.Responses weighted by each panelist’s confidence.

Source: Chicago Booth’s Kent A. Clark Center for Global Markets

When looking at experts’ responses, what clearly stands out is the high level of uncertainty on the future scenarios under AI. Half of the experts from the US and more than a third of European specialists are uncertain about the effects of AI on high-skilled workers and on inequality. The rest tend to be more skeptical about the earning potential of high skilled workers, but also on the possibility of AI shrinking income inequality. 

Key Highlights from PEP´s Survey

Our survey turned to a pool of more than 70 PEP economists, that mainly study the African Region and are concentrated in the Development Economics field. Most of them have a Doctoral Degree. We replicated the original survey questions, but differentiated the potential impact on developed and developing countries, and on high and low skilled workers. The following figures portray some of the survey´s results.

Figure 3. Use of artificial intelligence over the next ten years will have a negative impact on the earnings potential of substantial numbers of high-skilled workers in: (a) developed countries (b) developing countries.

Source: PEP.

Figure 4. Use of artificial intelligence over the next ten years will have a negative impact on the earnings potential of substantial numbers of low-skilled workers in: (a) developed countries (b) developing countries.

Source: PEP.

Even though the “uncertain”(or “neither agree nor disagree”) option results are lower than in the developing country survey counterparty, the “agree” and “disagree” options are more evenly distributed. The implications are therefore the same: opinions on AI’s effects on the earning of high and low skilled workers vary greatly, which places us a in a scenario of high uncertainty. On the other hand, specialists tend to agree slightly more on the fact that AI will have a negative effect on low-skilled workers´ incomes than on high-skilled workers´ incomes.

Figure 5.Use of artificial intelligence over the next ten years is likely to have a measurable impact in increasing income inequality in: (a) developed countries; (b) developing countries; and (c) between both.

Source: PEP.

PEP specialists tend to be more in agreement regarding the effects of AI in increasing inequality in both developed and developing countries, but also between both.

Think in scenarios, not certainties

The results of both surveys show large levels of uncertainty and discrepancy among specialists regarding the effects of AI on labor markets. As stated by Ethan Mollick, the large confidence intervals for predictions about AI should lead us to think in scenarios, not certainties. 

Nevertheless, what we do know for certain is that developing countries still lag in some crucial aspects needed for harnessing the benefits of AI, mainly regarding skills and infrastructure. As highlighted by PEP's unique survey of development economists, this gap creates particular challenges for developing countries. While the survey revealed high uncertainty about AI's specific impacts, itshowedstrongagreementamongexpertsaboutincreasinginequalitybothwithin and between developed and developing countries. This underscores why these countries   their population to navigate a fast-moving tech landscape. Developing countries should therefore prioritize the development of reliable and affordable internet connectivity and access to digital devices. Regarding skills, not only should developing countries enhance education and training programs that emphasize skills that complement AI, such as critical thinking, creativity, and emotional intelligence, but also tackle previous educational debts (low coverage, poor quality, scarce funding). As our survey respondents emphasized, addressing these fundamental challenges is crucial for ensuring that AI's benefits can be more equitably distributed and that technological advancement doesn't further widen the global development gap.