In 1814 Pierre-Simon Laplace described “intelligence” that, given state of all particles in the world and all the forces affecting them, could analyze these data and infer single formula for predicting all past and future events. This intelligence was later called “Laplace’s demon” and referred to as a first mentioning of scientific determinism. What is most often ignored is that Laplace pointed out to our practical inability to reach the state of this intelligence and thus need in probabilistic description of studied processes.
Probability theory, as well as its later derivatives such as statistics, machine learning and related fields, aim to provide tools for better reasoning in conditions of uncertainty. It has two main practical advantages. First, it helps to better understand world phenomena. Laplace, for example, refers to a comet of 1456 with unusually long tail that “spread terror through Europe”. Knowing true reasons for such events eliminates unreasonable fears and expectations, thus helping to make better decisions. Second, probabilistic models provide formal apparatus for automatic inference and forecasting. Be it text classification, face recognition or computational advertising, statistical methods bring best of two worlds: human intelligence and computer speed.
It’s also interesting how theory of chance describes progress in its own field. If infinite monkeys given infinite time and a typewriter create Hamlet almost surely, does it mean that infinite researchers of infinite generations will eventually invent Laplace’s demon?
[1]: http://bayes.wustl.edu/Manual/laplace_A_philosophical_essay_on_probabilities.pdf