This resource is no longer available
Humility in AI means having the tools to recognize and act on situations in which a model may be less confident of a decision. This may happen because the prediction falls near the decision boundary, or could be caused by either the quality of the data the model was trained on or the quality of the data the model is now being asked to score. A lack of diversity in your training data will leave a model ill-prepared to make confident predictions on those less-seen characteristics. Outliers or outright anomalies in scoring data will also cause model performance to suffer.