Causal Inference and Statistical Learning!
ML beyond Curve Fitting: An Intro to Causal Inference and do-Calculus. you can attempt to perform full causal discovery: attempting to infer the causal model or at least aspects of it, from empirical data.
But the bottom line is: a full causal model is a form of prior knowledge that you have to add to your analysis in order to get answers to causal questions without actually carrying out interventions.
Learning theory and Algorithmic information theory
Causal inference using the algorithmic Markov condition
Causal Markov condition for submodular information measures
Probality-free causal inference via the Algorithmic Markov Condition