This collection helps humanitarian teams describe and compare specific uses of AI. Start with the work someone needs to do, then identify the AI capabilities, permissions, people affected and evidence required for a decision.
The collection continues Thomas Byrnes’s earlier humanitarian AI research. Its classification and risk foundations are the DCI/GIZ taxonomy and co-authored risk framework. SAFE AI adds humanitarian governance guidance.
Find your starting point
Begin with the humanitarian AI definition and the concept map. Read the classification method to understand the capability and function codes. Use the risk page to see how the assessments remain separate. The translation and complaints examples show how to apply the method.
What the collection establishes
These are draft reference entries and invented worked examples. They are not operational assessments or approvals. The original authors remain credited. AidGPT’s proposed categories and applications are identified as adaptations.
Sources
A. A Taxonomy for AI in Social Protection: Defining Capabilities and Application Areas
F. SAFE AI Governance Framework