How can the area under the receiver operating characteristic curve (AUC-ROC) be interpreted in the context of model performance, and what are the limitations of using this metric alone for assessing the effectiveness of a binary classifier?

How can the area under the receiver operating characteristic curve (AUC-ROC) be interpreted in the context of model performance, and what are the limitations of using this metric alone for assessing the effectiveness of a binary classifier?