Getting the foundations right: How South Asia can make AI work for growth

Two forces are currently reshaping the global economy. The first is high energy prices, resulting from the conflict in the Middle East. South Asia is particularly vulnerable. It is one of the most energy-dependent regions in the world, and it relies more heavily on shipments through the Strait of Hormuz than other regions. The second is the rapid advance of artificial intelligence (AI). Countries that produce AI-related goods have been propelled forward by rising exports and investment. So far, South Asia has largely missed these AI gains. Yet the region continues to grow faster than other regions, and growth is projected to remain steady near 7 percent in the coming years. What explains this continuing resilience? A key contributor has been the long-term structural reforms that South Asian governments have implemented steadily over the past decade. In India, this has involved consolidating the labor code, rationalizing the goods and services tax, adopting free-trade agreements, and increasing spending on public infrastructure. In Sri Lanka, it has been mobilizing revenue, transforming utilities, and rationalizing tariffs. In Bhutan, it has been liberalizing foreign direct investment. These reforms take time to bear fruit. Their impact is gradual and cumulative. But the benefits can now be seen in South Asia’s robust domestic demand, despite slowing growth in the rest of the world. South Asia needs to sustain this momentum and invest in new drivers of growth. Just as today’s resilience is a product of yesterday’s reforms, future resilience will be a product of the reforms being undertaken today. This is where South Asia can harness the opportunities provided by AI. The potential is there. Far more South Asian workers stand to gain from AI than to lose to it.Only 7 percent of jobs in the region combine high AI exposure with low human complementarity, compared to 15 percent elsewhere. Currently, only about one in four Indian firms use AI—compared to one in two in the United States—and Indian firms tend to do so in less sophisticated ways. But adoption is accelerating fast, and for good reason: positions requiring AI skills already command a wage premium of nearly 30 percent. South Asia is also already being pulled into the AI economy through its global supply chains. American firms that have adopted AI heavily are deepening their ties with South Asian suppliers rather than cutting them. Following the launch of ChatGPT, a one-standard-deviation increase in a U.S. firm's AI exposure has been associated with a 27 percent larger rise in its customer links with South Asian firms, and that effect more than doubles when the South Asian supplier is itself in a sector that is highly exposed to AI. In short,AI is not replacing South Asia in global value chains. Instead, it is deepening linkages. The technology is also beginning to show up in the daily work of doctors, farmers, and teachers. In Telangana, India, AI-based weather forecasts are saving smallholder farmers up to $560 per season by improving planting decisions. In Bangladesh, AI-assisted retinal screening has increased the number of patients checked for diabetic complications by 40 percent per day. These are early signals of what broad AI adoption could do for productivity and living standards. But AI cannot outrun the foundations on which it depends. To fully reap the benefits of AI, countries need to address foundational gaps by enriching workforce skills, strengthening the business-enabling environment, and improving physical and digital infrastructure. Governments are sometimes tempted by flagship projects — building frontier AI models or funding showcase data centers. Our evidence points elsewhere. South Asian firms themselves identify digital infrastructure and tertiary education as the two biggest constraints on AI adoption. In South Asia, reliable electricity, strong education systems, and digital connectivity will do more to spread AI For most of South Asia, the gains lie in adopting and adapting existing tools to local conditions, languages, and connectivity constraints. This means emphasizing "small AI": practical applications adapted to specific tasks and local contexts, running on basic devices in low-bandwidth environments. These applications can increase productivity and incomes in the sectors that employ the most people. South Asia region has demonstrated remarkable resilience in the face of persistent shocks. By harnessing AI’s potential and investing in foundational reforms, the region can ensure continued growth and jobs for the future. Read more in our latest South Asia Economic Update: Adopting AI for Growth.