Executive Summary
Artificial intelligence is increasingly used by governments and social sector organizations in low- and middle-income countries. From helping farmers optimize crop yields in Malawi to delivering personalized adaptive learning content to parents and daycare workers in India, organizations are deploying predictive and generative AI across a wide range of use cases.
Implementers often deploy AI to improve efficiency, speed, and scale, using AI to deliver higher-quality services to more people while reducing costs. These are undoubtedly important outcomes – and in GiveDirectly’s work, we’ve found that using AI tools can help us more accurately and quickly identify people in need during a crisis, and send payments up to 4-8x faster than we would have by relying on in-person operations.
However, a range of other outcomes are essential to the long-term sustained success of development and humanitarian programs–such as including the most vulnerable, treating program participants with dignity, and enabling informed consent around participation. Prioritizing these outcomes is critical for ensuring that development and humanitarian efforts achieve long-term objectives of improving human welfare while reducing inequality and social marginalization.
