Why We Need AI in Humanitarian Aid: Efficiency in Crisis Response

Last week at the Eesti Pagulasabi / Estonian Refugee Council International Humanitarian Aid event in Tallinn (November 7, 2024), I joined an enlightening panel discussion with Karin Maasel Maasel (CEO, Data Friendly Space) and Robbie Phillips (Humanitarian Science and Technology Lead, FCDO), expertly moderated by Kristen Davis (CEO, CinqC). I want to share why AI's efficiency gains are becoming critical for our sector.

The Problem: A Sector Under Pressure

Let me start with a stark reality I shared during the panel: in 2016, global humanitarian funding averaged $160 per person in need. Today, that figure has plummeted to $60. This dramatic decline means we must fundamentally rethink how we operate. As I explained, humanitarian agencies now face an impossible equation - increasing needs, decreasing resources, and growing complexity of operations.

In my 14 years working across national NGOs, the Red Cross Movement, and the UN, I've never seen humanitarian agencies so overstressed. We're trying to make complex operations work with diminishing resources, while the scale and complexity of crises continue to grow.

AI's Efficiency Multipliers:

1. Rapid Data Processing During the panel, I shared a recent example from my work supporting the North West Syria response Cash Working Group. A local staff member used AI to analyze a Telegram video of arriving refugees from Lebanon. Within minutes, AI estimated approximately 2,000 people (final count: 2,300) - a task that would have taken much longer through traditional counting methods. This wasn't just faster - it was actionable intelligence that enabled immediate response planning.
2. Force Multiplication of Limited Teams During the Turkish earthquake response, I shared how the IFRC's three-person information management team leveraged AI to develop Python scripts for market analysis. They used AI not to analyze data, but to rapidly develop tools that could scrape Turkish websites, pull information from national statistics, and create visualizations. This allowed a tiny team to provide critical market functionality data within hours - information that would have been useless if it took weeks to compile.
3. Processing the Previously Impossible One of the most powerful efficiency gains I highlighted was in processing vast amounts of previously untouched data. Meeting minutes, complaints mechanisms, and community feedback that often went unanalyzed due to resource constraints can now be processed and acted upon. As I noted during the panel, "These people are bearing witness, and we're finally able to listen."
4. Automated but Accountable A key point Karin and I discussed was how AI enables automation while maintaining transparency. Their Gannet system, for instance, shows exactly which sources inform its analyses - something general AI tools often lack. This means we can gain efficiency without losing accountability.

Looking Forward

The efficiency gains from AI aren't just about doing things faster - they're about enabling our sector to maintain its effectiveness despite severe resource constraints. As I emphasized in Tallinn, we need these tools not to replace human judgment, but to ensure our limited human resources can focus on decision-making and community engagement rather than data processing.

What's your experience with AI efficiency gains in humanitarian work? Have you seen similar transformative effects in your operations?

Please continue the discussion at aidgpt.org

#HumanitarianAid #ArtificialIntelligence #Innovation #Development #HumanitarianTech #Aid #EstonianRefugeeCouncil #Tallinn