Prediction is not the same as understanding cause and effect. Kaushik Rajan highlights how ignoring causality can turn high-performing models into risky ones. https://towardsdatascience.com/causal-inference-is-eating-machine-learning/