Why machine learning struggles with causality

Until now, machine learning has neglected a full integration of causality, and this paper argues that it would indeed benefit from integrating causal concepts.