Machine learning and AI are sweeping through all aspects of public and private life, but one key challenge is how to move beyond simple correlations and predictions to causal knowledge that can guide action, policy, and plans. Over the past 30 years, a number of algorithms for learning and using causal knowledge (even from purely observational data) have been developed. In this talk, I will first describe the state of the art in causal discovery and reasoning methods, many of which have been developed at CMU. I will then outline various natural uses of these algorithms in (national) security and defense contexts, drawing from examples of research and applications conducted with the SEI.
See the slides.