ISCN Global Mixer 46: Tectonic shift in urban disaster response

Tectonic shift in urban disaster response: 

How GeoAI enables effective response when disaster strikes cities

fAIr is an open-source, AI-assisted mapping initiative for disaster response provided by the Humanitarian OpenStreetMap Team (HOT). It works with satellite and aerial imagery to use AI for building footprints, density estimates, and damage assessments in cities - all as open code and open data. Crucially, human validation is built in through MapSwipe, where volunteers verify AI predictions before they reach responders.

This proved critical in the June 2026 Venezuela earthquake. Within 24 hours, using GeoAI to estimate building density and damage layers across Caracas and La Guaira, areas missing most open map data are identified. Volunteers validated each prediction to ensure hallucinations were addressed, and the open data was used by people on ground to plan aid and recovery.

Key Takeaways for Municipal Decision-Makers:

• When disasters strike, timely information about where people may be affected is essential for effective response and informed decision-making. HOT provides access to emerging technologies that help cities and local authorities improve situational awareness and strengthen disaster response.
• fAIr is an open, community-driven platform that connects GeoAI models with real-world mapping needs. It enables municipalities and humanitarian actors to benefit from AI-assisted mapping without requiring in-house AI or machine learning expertise. 
• The Venezuelan case demonstrates how AI-based analyses and predictions can be validated and verified using swarm intelligence of contributing volunteers on the ground – and how local authorities benefit from it.
• MapSwipe integrates a Human-in-the-Loop (HITL) approach by enabling volunteers to validate spatial data quickly and at scale. This provides cities with an additional layer of confidence when using AI-generated mapping products for disaster risk management.

• There is no one-size-fits-all approach to AI-assisted mapping. Every city has different data gaps, risks, and operational needs. HOT supports municipalities in identifying where AI can provide the greatest value and in selecting open AI models that best match their local disaster risk contexts.

HOT has issued an open call for earth observation ai models: the global GeoAI community is invited to contribute their open source GeoAI models to the platform.

The slides presented can be downloaded here

Interested in collaborating with Pete and Kshitij?  Reach out to us via ISCN@giz.de 

Interested in learning more about Kshitij’s and Pete’s work?

• fAIR
• Home - HOT Website
• Tools & Resources - HOT Website
• Humanitarian OpenStreetMap Team on GitHub
• Online Course by HOT: Humanitarian Use of OSM Data