We provide a faster algorithm for generic identification in tree-shaped linear structural causal models.

Statistical prediction may be sufficient to drive the emergence of internal causal models and causal inference capacities in deep neural networks.