--- aliases: en: - capability eval - capability evaluations concept_id: capability_evaluation preferred: en: capability evaluation ja: 能力評価 relations: relations: broader: - evaluation prerequisite: - capability - capability_elicitation - evaluation related: - control_evaluation status: active --- # capability evaluation An evaluation designed to measure whether an AI system can perform specified tasks or behaviors under an elicitation protocol, rather than whether it tends or chooses to perform them in ordinary use. ## broader - [evaluation](../terminology/evaluation.txt) ## prerequisite - [capability](../terminology/capability.txt) - [capability elicitation](../terminology/capability_elicitation.txt) - [evaluation](../terminology/evaluation.txt) ## related - [control evaluation](../terminology/control_evaluation.txt) [Corpus index](../index.txt)