GiveDirectly and Google.org are using AI mapping and satellite imagery to find low-income hurricane survivors in Puerto Rico and Florida and send cash payments
Storms worsen inequality. These $700 cash payments can help lessen the blow. — After a disaster like Hurricane Ian … Tweets: @themayadecline Tweets: Maya / @themayadecline : mapping project that @hanshengchia @alexnawar the google pals and i worked on, being used to help people! https://www.fastcompany.com/ ...
Context & Ripple Effects
GiveDirectly's hurricane-response work with Google.org is one node in a longer arc of Google philanthropy aimed at AI for social impact, going back to its $25M AI impact grants across a dozen countries. The nonprofit had already been using Google's Earth Engine satellite imagery and analysis tools for environmental monitoring before applying the same imagery stack to targeting survivors of Hurricane Ian in Puerto Rico and Florida.
What makes this story notable is where it points: three years later, GiveDirectly is moving from reactive payouts to pre-positioned aid, planning to use Google's AI-based Flood Hub to send early aid to at-risk families in Bangladesh. That shift rests on Google's forecasting progress — AI flood warnings accurate up to seven days ahead across more than 80 countries, per Google Research's own paper, and an [[a:886772|AI cyclone model now being tested with the US National Hurricane Center through Weather Lab]].
First-order effects
- Low-income households in Puerto Rico and Florida hit by Hurricane Ian receive $700 cash payments without having to navigate application processes, because AI mapping of satellite imagery identifies who qualifies.
- GiveDirectly gains a targeting method that scales beyond traditional door-to-door or sign-up-based disaster relief, with Google.org supplying the imagery and model infrastructure.
Second-order effects
- Other humanitarian funders face pressure to match machine-speed targeting — once aid can be routed by satellite-derived income and damage estimates, slower survey-based disbursement looks like a competitive disadvantage for donations.
- Google's weather-forecasting assets become inputs to the aid pipeline itself: the Flood Hub and cyclone models turn prediction products into distribution channels for GiveDirectly-style cash programs.
Third-order effects
- If the pattern holds, disaster philanthropy restructures around forecast-triggered payments — aid released before or at landfall based on AI predictions rather than after damage assessments — with tech companies' geospatial and meteorological models becoming core humanitarian infrastructure.
- That consolidation raises a governance question the corpus does not resolve: which organizations audit the models that decide who gets help, as private-sector forecasting systems effectively gatekeep public and charitable relief.
The trend: Disaster relief is shifting from post-storm assessment to AI-predicted, pre-positioned cash aid, with Google's forecasting and imagery stack becoming the shared infrastructure nonprofits build on.