Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead

The two main takeaways from this paper: firstly, a sharpening of my understanding of the difference between explainability and interpretability, and why the former may be problematic; and secondly some great pointers to techniques for creating truly interpretable models.