AI in Humanitarian Aid and Refugee Response Quiz

Answers and explanations appear as you take the quiz.

1. What kinds of data did Togo's Novissi programme combine to target cash transfers?
   • Hospital records and school attendance
   • Satellite imagery and mobile phone usage patterns
   • Tax returns and property deeds
   • Social media posts and search histories
2. What is a known limitation of phone-based aid targeting?
   • It cannot send money electronically
   • It only works in wealthy countries
   • It can miss people who do not have phones
   • It requires every recipient to have a bank account
3. Which agency's tool, HungerMap LIVE, nowcasts food insecurity where recent surveys are missing?
   • The World Food Programme
   • The World Health Organization
   • The International Monetary Fund
   • UNESCO
4. Why did Human Rights Watch raise concerns about Rohingya refugee data in 2021?
   • The data was deleted before aid could be delivered
   • The data was sold to advertisers
   • Biometric scanners were found to be inaccurate
   • Data was shared with Myanmar without adequately informed consent
5. When evaluating an aid-targeting model, why should teams report exclusion errors separately?
   • Exclusion errors are always zero for machine learning models
   • They show how many genuinely poor people were missed
   • They measure how fast payments are sent
   • They are required to calculate satellite resolution
6. What does FEWS NET use to classify the severity of food insecurity?
   • A three-colour traffic light scale
   • A percentage of GDP
   • The five-phase IPC scale
   • A 1-to-10 risk score from a single model
7. Which statement reflects the guide's view of consent in humanitarian data collection?
   • Consent is hard to make meaningful when aid depends on agreeing
   • Consent is unnecessary if data is encrypted
   • Consent is only needed for children
   • Consent is guaranteed by using biometrics
8. How are satellite poverty-mapping models typically trained?
   • By reading text written on rooftops
   • By predicting wealth measured in household surveys from image features
   • By counting the number of cars in each photo only
   • By copying government census maps directly