Oct 2, 10:01 AM
The World Bank’s Unfashionable Case for a Smaller AI
Rather than treating artificial intelligence as a clean route to more concentration and more hype, the World Bank argues that tailored systems could help farmers, schools and clinics in poorer economies.
Artificial intelligence tends to arrive in public debate wearing two costumes: miracle and menace. The World Bank is trying a more prosaic outfit. Its argument is that AI should not be judged only by the damage it might do in rich economies, but by whether it can be made useful in places where basic services still matter more than digital theatrics.
Paschal Donohoe, managing director and chief knowledge officer at the World Bank Group, says the technology already has concrete uses in farming, education and healthcare. In agriculture, he points to crop productivity and to the rather less glamorous business of helping farmers decide what to plant and when to harvest. He cites an example from India, where small rice farmers received better advice on planting and harvesting through the use of artificial intelligence, with the aim of improving productivity and the income they earn.
The World Bank’s preferred answer is what it calls small AI. The phrase sounds modest because it is meant to be. Donohoe says such systems need to be tailored to local conditions, able to run offline and capable of working where connectivity or electricity is weak. That may not sound like the sort of thing that excites conference panels, but it is closer to the reality of development than the usual parade of grand promises.
The bank’s broader point is that development agencies cannot afford to discuss AI in isolation from the old, stubborn questions of poverty, clean water, healthcare and education. Donohoe argues that the task is to recognise the change AI is bringing without losing sight of those basic needs. The World Bank’s case is simple enough: with the right policies, and with systems designed for the environments in which they are actually used, AI could help narrow gaps rather than widen them.
That is a useful correction to the usual habit of treating technology as either salvation or catastrophe. In practice, the question is more awkward and more interesting: who gets to benefit from AI, and under what conditions? If the answer is to include developing economies, then the machine will have to be smaller, humbler and a good deal less dependent on perfect infrastructure. Which, for once, may be exactly the point.
Written by Freya Stensrud freya.stensrud@alpineweekly.com
