Definition
arrangement through which responsible people can observe, challenge, intervene in or stop a humanitarian AI use during its operation
AidGPT working definition. Source basis below.
Notes to entry
Note 1 to entry: Oversight needs actual access, time, competence and authority. Merely naming a person in a process does not establish effective control.
Note 2 to entry: ISO/IEC TS 8200:2024 addresses controllability, including state observability and control transfer. NIST AI RMF 1.0 calls for defined and documented human oversight. This AidGPT definition applies those ideas to a particular humanitarian AI use; it is not quoted from either source.
Source basis
AidGPT working definition. These sources informed the concept and its scope. They did not issue or approve this wording.
- DCI AI Hub, AI Hub Risk Assessment Framework for Social Protection (Sector risk framework)
- Core Humanitarian Standard on Quality and Accountability, 2024 (Humanitarian sector standard)
- NIST AI Risk Management Framework 1.0, Core (National risk framework)
- ISO/IEC TS 8200:2024, Controllability of automated AI systems (Technical Specification)