An organization's AI risk framework includes a requirement for 'AI system resilience.' Which risk scenario BEST tests AI system resilience?
Select an answer to reveal the explanation.
Short Explanation
Here's the deal — resilience testing checks how your AI system behaves when things go sideways. You want to see if it keeps acceptable performance and fails safely under noisy inputs, missing data, and unexpected operational conditions.
Full Explanation
Resilience testing assesses how a system behaves under adverse or unexpected conditions, not just normal operation. For AI systems, this includes testing performance under noisy inputs, partial data outages, edge cases, and adversarial perturbations, as well as ensuring graceful degradation rather than catastrophic failure. Evaluating target AUC on a holdout test set tests normal-condition performance. Confirming pipeline completion within the batch window is an operational concern. Verifying complete signed documentation is a governance check.