Scruitability is he potential for an algorithm or model to be undersyood and interpretted by a human. Soem representations, such as decsion trees are more naturally scrutible, at least by experts. Other, such as neural networks are opaque and hence the need for techniques in explainable AI to expose aspects of these blaxk-box models. Scrutability is important to ensure legal compliance and xpose potential bias in the model.
Used on Chap. 18: page 439