Why AI is not a quick leapfrog for many developing countries

In short: A Financial Times interview argues that AI can help developing countries, but it usually works best when basic systems and delivery channels are already in place.

What's going on

A Financial Times newsletter spoke with Rose Mutiso, executive director of the African Tech Futures Lab, about how AI is used in developing countries. One example was in India, where about 38 million farmers received AI-powered forecasts for when the monsoon rains would start.

The forecast combined two AI models and gave farmers about four weeks’ notice. It even spotted a false start to the monsoon, which helped farmers decide what to plant and when.

Mutiso said this was not a “just add AI” story. It worked because India had decades of global climate records, around 100 years of local rainfall data, and a long-running system for sending advice by SMS in local languages. In other words, AI acted like a powerful amplifier, not a magic replacement for missing basics.

The interview also pushes back on the common idea that AI will help countries “leapfrog” (skip older steps, like how many places skipped landlines and went straight to mobile phones). Mutiso said AI is different because it needs a lot of physical resources like reliable electricity and data centers (buildings full of computers). She noted Africa has around 1 percent of global data center capacity.

What to watch

Mutiso suggested focusing on “sequencing”, meaning picking specific problems where AI can help now, while slowly building the supporting data, power, and institutions. She also warned about job risks, including poor working conditions for data labeling work, where people tag text or images so AI can learn (like putting name labels on thousands of folders).

Source: Financial Times