Interpretability refers to the ease with which a human can understand and trace the decision-making process of a machine learning model. For example, a decision tree is naturally interpretable as its structure explicitly shows how input features lead to predictions. In contrast, deep neural networks are often seen as black boxes due to their complexity, making interpretability challenging but crucial for trust and transparency.