Abstract for: Is Observed Data Adequate to Automate the Construction of Causal Models?

System dynamics models are built on a set of cause-and-effect assumptions representing the understanding of the causal relationships in a real system. Can machines automatically build system dynamics models by inferring causal assumptions directly from observed data? In this paper, we provide a review of existing works on automated model building and critically discuss their strength and limits. Relating the question to recent studies of causal inference, we conclude that state-of-the-art techniques are still inadequate to automatically infer causal relations from observed data, that the inadequacy has to do with different understandings of causality, and that the relation between causal model and empirical data deserves a thorough consideration as we introduce more data science and artificial intelligence techniques into the system dynamics field.