The global growth slowdown: AI to the rescue?

Artificial intelligence (AI) is advancing rapidly at a moment when the global economy badly needs a new source of growth. The stakes are especially high in emerging market and developing economies (EMDEs), where growth has slowed to a two-decade low just as about 1.2 billion young people are set to reach working age between 2025 and 2035, likely their largest youth cohort ever. These economies need faster growth and more productive jobs. With the right complementary policies and investments, AI could help EMDEs achieve both, extending its benefits beyond the advanced economies already best equipped to use it.
 

A protracted growth slowdown

Global potential growth—the pace an economy can sustain without generating inflationary pressures—has fallen from 3.6 percent a year in the 2000s to 2.8 percent in the 2010s. It is projected to average about 2.2 percent in the 2020s. The decline has been steeper in EMDEs, where potential growth is projected to fall from 5.9 percent in the 2000s to 4.1 percent this decade (Figure 1). Weaker investment, slower labor force growth, and declining productivity growth have all contributed to this broad-based slowdown around the world

Enter the era of AI

AI is already spurring investment, particularly in advanced economies, while rapidly growing trade in AI-related goods is spreading some of that demand across borders. Over time, however, its greatest promise lies in raising productivity. Early evidence is encouraging in some settings: for example, customer-support agents in the Philippines who had access to AI assistance increased their productivity by 15 percent, on average. Other studies find even larger gains in certain tasks, often for less-experienced workers.

The leap from task-level gains to economy-wide growth is much harder. Firms must adopt AI, reorganize work, train employees, and identify tasks where AI can improve productivity. Much of the evidence still comes from advanced economies. Some expert users have seen little benefit, and many tasks may prove difficult to automate or improve. As a result, estimates of AI’s contribution to annual productivity growth range from about 0.07 percentage point in cautious assessments to as much as 1 percentage point in optimistic ones (Figure 2).

This wide range of estimates reflects uncertainty about the share of tasks that can benefit from AI, the productivity gains within those tasks, and the pace of adoption across firms, among other assumptions. A larger share of AI-compatible tasks or faster adoption implies greater economy-wide gains. The evidence so far comes largely from early adopters, whose experience may not generalize broadly.

Earlier technologies offer perspective and a reminder that benefits take time to spread. Electrification may eventually have added about 0.9 percentage point to annual productivity growth in Europe, but factories had to be redesigned and supporting infrastructure installed first. The information and communications technology revolution contributed about 0.5 percentage point to annual productivity growth in the United States. AI’s eventual impact remains uncertain, and its broader gains are likely to emerge gradually.

Could AI reverse the slowdown?

What could AI-driven productivity gains mean for growth? We explored three illustrative scenarios in which AI raises productivity growth from 2026 through 2039. Two scenarios assume annual increases of 0.6 and 1 percentage point, respectively. These are the median and high-end estimates of the impact of AI across studies. The third scenario considers a transformative outcome comparable to the strongest ten-year productivity growth performance recorded across economies. In this scenario, global productivity growth in the 2030s would more than triple relative to the baseline, rising from 0.8 percent to 2.7 percent per year.

All three scenarios lift global potential growth, but by very different amounts. In the transformative scenario, potential growth in the 2030s would exceed its average in the 2000s by 0.5 percentage point per year, making the 2030s possibly the strongest decade for global growth since the 1970s. The other two scenarios deliver additional growth without restoring global growth to its 2000s average (Figure 3).

These scenarios are illustrations, not forecasts. They abstract from the time and investment needed to adopt AI, changes in labor markets, and other forces that could strengthen or weaken its effects. They also assume similar productivity gains in advanced economies and EMDEs, an especially optimistic assumption for the latter.

Preparing EMDEs for AI

EMDEs are less prepared, on average, to adopt AI. They have lower levels of internet use, digital skills, and overall AI preparedness than advanced economies (Figure 4). As of mid-2025, they hosted less than a quarter of global data-center capacity; low-income countries hosted less than 0.1 percent. In 2022, only one in four people in low-income countries used the internet. These gaps could limit the productivity gains EMDEs achieve relative to those illustrated in our scenarios.

Even as they address these gaps, EMDEs can pursue practical applications adapted to local needs and circumstances. Applications that run on mobile phones with limited connectivity, use local languages and data, and are inexpensive to deploy could extend AI’s benefits well beyond large firms and major cities. Such “small AI” applications could help farmers identify crop diseases, provide customized tutoring, or support health workers in remote areas.

A. Panel shows scores on the AI Preparedness Index (AIPI) as of 2023 by Cazzaniga et al. (2024), covering 174 economies. Higher values represent more favorable AI preparedness. Country groups are as defined by the International Monetary Fund and aggregated using averages.
B. Bars show the median share of individuals using the internet in advanced economies and EMDEs. Sample includes 37 advanced economies and 154 EMDEs using the latest available data for each economy.

Looking forward

AI could offer a path to reversing the prolonged slowdown in global growth, but only if productivity gains are large, persistent, and widely diffused. For EMDEs, realizing this promise of AI requires investing in reliable electricity, data, digital connectivity, and skills, while supporting workers as tasks and occupations change. It also requires an enabling environment for businesses to grow and for an AI ecosystem to develop and adapt tools to local needs. Mobilizing private capital at scale will be essential to finance firms and ease constraints on growth, helping translate AI’s gains into the productive jobs their growing workforces need. These policy and investment choices will determine whether AI becomes a broad-based driver of global growth or delivers gains concentrated mainly in advanced economies.

Source : World Bank