What we will measure
Low-income contexts have little real-time measurement, so it will need to be constructed. We are working on new projects that measure AI access, usage, local usefulness, and economic effects. Stay tuned for more.
Help build the evidence →Questions in this area
How do the world’s poor use AI?
What we've found
In a project in Sierra Leone, teachers submitted 40,350 queries to an AI chatbot in WhatsApp over 17 months. More teachers used the chatbot than web search for teaching assistance each month.
What we're asking
We're exploring who else is currently using AI, with a particular eye to less resourced populations. Has AI reached only people of certain income levels, education levels, and regions? What do they use it for? Do only speakers of certain languages use AI?
What are the bottlenecks to benefitting from AI?
What we've found
SMS charges may put AI out of reach for roughly 1 billion basic-phone users. Yet AI can use scarce bandwidth much more efficiently than web pages.
What we're asking
What most constrains the benefits of AI: connectivity, devices, costs, language gaps, local data, education? How can we use data to measure these gaps? What policies and markets can best fill them?
How might AI transform economies?
What we’re asking
AI could have transformative economic effects. How could machine intelligence reshape economic opportunities for people in low- and middle-income countries? Will AI leave the developing world behind? How should policy respond?