To reduce the risk of harmful automated decisions, an organisation ensures a competent person can review and, where needed, override the AI system output. This control is:
- Afully autonomous operation
- Bhuman oversight (human-in-the-loop)
- Cremoving all monitoring
- Ddisabling audit logging
Human oversight (human-in-the-loop) ensures a competent person can review and override AI outputs, reducing the risk of harmful automated decisions.
Why each option is right or wrong.
- B (correct). A person reviewing and able to override the system is human oversight.
- A. Full autonomy removes the oversight control.
- C. Removing monitoring increases, not reduces, risk.
- D. Disabling logging weakens governance.
Reference: IAPP AIGP, Governing AI deployment.