Reading https://www.thereporterethiopia.com/52086/ I like the emphasis on developing the infrastructure around AI. But that part about predictions... Everyone expects a perfect system in these big vision statements. "AI will do magic." Even if you can build a 99% accurate prediction, how do you tell when it's the 1% or the 99% with what you're looking at? Maybe it doesn't matter? In which case, why not? To sidestep dealing with all of those uncertainties. The question to ask before building anything is, how can this be useful and valuable even when it sucks? How can we design the human and machine system such that responders acting on imperfect predictions still build resilience to the communities regardless of accuracy? Prove value for a broken system first, before assuming you can build a perfect system. But, you don't excite funders talking about crappy systems, do you?