Learn about the key differences between post-hoc explainability and inherently interpretable models in this 24-minute data science lecture that explores the select-then-predict framework. Dive into Lei et al.'s 2016 research, understand the concept of differentiable binary variables, and examine the pipeline approach to model interpretation. Discover the implications of Trojan explanations and their impact on model interpretability while gaining practical insights into creating more transparent and explainable machine learning models.