Emerging economies need to use AI differently

The objective of emerging economies should be to maximise augmentation, not automation, through AI adoption

The most successful member of my St Stephen’s batch, World Bank president Ajay Banga (we all take advantage of the reflected glory!) recently made the important observation in a panel discussion: Generative AI (GenAI) will impact emerging markets very differently from how it is being seen to affect advanced economies. Push that argument a little further, and an intriguing possibility emerges. Can emerging economies that make the right choices use GenAI to narrow their gap with advanced economies? At first glance, that sounds counterintuitive.

AI does not need to replace the work done by a farmer, construction worker, or nurse. It can improve decisions related to such work. (Shutterstock)

GenAI is extraordinarily capital-intensive. It requires advanced chips, enormous computing capacity, reliable electricity, data centres, data, large language models, and, increasingly, sophisticated foundation models. On almost every one of these dimensions, emerging economies start at a big disadvantage. The scale of AI investment in the US (in trillions of dollars, more than the GDP of many emerging economies) alone makes the point. The most advanced chips are scarce and expensive. Training and operating frontier models requires vast amounts of capital and energy. The leading models are being developed by a relatively small number of companies concentrated largely in the US, with China also investing heavily.

Much of the debate about AI in emerging economies simply imports the concerns of advanced countries, such as job displacement by automation. Yet, the structure of their economies is fundamentally different. For emerging markets, the bigger opportunity may not be to produce AI, but to use it to gain access to something they have historically lacked — expertise at low marginal cost.

That distinction matters enormously. Think of a dairy farmer who rarely sees a veterinarian; a nurse or community health worker operating far from a specialist physician; or a small entrepreneur who has little access to an accountant, lawyer, marketing adviser or management consultant. GenAI can potentially place elements of all these capabilities in the hands of ordinary workers, in their own language and at very low marginal cost.

That could make its economic impact in emerging economies quite different from that in advanced ones. The first fear surrounding AI is job loss. But fewer jobs in emerging economies are directly exposed to GenAI automation (between 4.5-11%). India alone has over 500 million people dependent on agriculture. In advanced economies, by contrast, employment is much more concentrated in clerical, administrative, professional and highly digitised occupations (according to ILO, around 30% of jobs) — precisely the areas most exposed to GenAI.

AI does not need to replace the physical work performed by a farmer, nurse, caregiver, technician or construction worker. It can improve the decisions surrounding that work and, therefore, raise productivity substantially. Consider the dairy farmer. AI cannot milk a cow. But it could help identify early signs of disease, recommend changes in feeding, improve breeding decisions and indicate when veterinary intervention is necessary. A farmer could photograph a damaged crop and receive advice on pests, irrigation, fertiliser or weather risks in a local language. The real innovation is not automating the farmer. It is giving the farmer something resembling an agronomist, veterinarian, and market adviser on demand.

The same logic may be even more powerful in health care. Many advanced economies are ageing rapidly. Japan, South Korea, China, France and Germany, all face growing pressures from ageing populations, while health care systems are already struggling to provide care, bedeviled as they are with workforce shortages and growing budget deficits. If AI displaces significant numbers of workers, consumption and labour-based tax revenues could come under further strain.

Emerging markets will get a big opportunity with younger, cheaper, and AI enabled workforces. Think nurses, community health workers and caregivers supported by AI could use better triage protocols, documentation, translation, medication checks and monitoring. In effect, AI could allow less highly trained workers to perform some tasks more reliably and productively.

Emerging economies also do not have to go first. AI adoption is likely to advance fastest where wages are highest, digital infrastructure strongest and employment most concentrated in AI exposed occupations. Developed economies will, therefore, confront the hardest questions earlier — automation, job redesign, reskilling, inequality and the distribution of productivity gains. The tax system itself will need reform.

Emerging economies can watch, learn, and make better-informed choices. Their objective should not be to maximise automation. It should be to maximise augmentation. That means directing AI towards sectors employing large numbers of people, developing applications in local languages, building voice-first interfaces for workers with limited digital literacy, investing in training, and designing regulations that distinguish clearly between decision support and decisions that must remain with qualified professionals.

Emerging economies may never match advanced economies chip for chip, data centre for data centre or dollar for dollar. They may not need to. The greater opportunity is to take intelligence created at enormous cost elsewhere and distribute it at marginal cost across millions of workers.

There is, of course, no guarantee that this will happen. Technology alone does not determine economic outcomes. Policy, infrastructure, education, business models and regulation will all matter.

But the strategic choice is clear. The internet democratised access to information. GenAI has the potential to democratise access to expertise. For advanced economies, AI may primarily be a technology of automation. For emerging economies, it could become a technology of capability. If that opportunity is seized, farmers could become better farmers, nurses’ better nurses, and small entrepreneurs’ better businesspeople. Unlike previous waves of innovation that entrenched the advantage of rich countries, GenAI may help bridge the gap. Yes, I have a dream too, that one day an Indian child may get the same opportunities as an American kid. What a world that would be!

Janmejaya Sinha is chairman (India), BCG. The views expressed are personal