The Book of Why
Judea Pearl is one of the fathers of modern Bayesian Networks which are pretty much used everywhere these days. So, in my pursuit to find a book talking about AI, I stumbled across this book. This book explains Judea’s latest contribution to computer science which is a mathematical approach to modeling causality. In the book he starts by explaining where the phrase “correlation does not equal causality” comes from. His argument is that with statistics you will never be able to define causation, because statistics does not have the language or framework to make such statements. As such statistics and big data can only go so far in their abilities to provide answers for our questions. Instead of approaching problems with the esoteric methods of regression, data stratification and trying to control for various values based on intuition, he proposes that we should try to approach problems more like human beings. He breaks the idea of causality into three rungs on a ladder. He proceeded to explain a sort of calculus that quantifies the effects that different causal relations have on the outcome of a situation.