The most interesting projects in AI are in population health, not Silicon Valley 

To use AI to fix problems in the developing world, researchers need to think differently.

Source: AI researchers need to approach developing world problems in a completely different way — Quartz