Bayes' Theorem describes how to update the probabilities of hypotheses when given new evidence. It underpins Bayesian inference, where prior beliefs are combined with observed data to produce posterior beliefs. Many machine learning methods, such as Naive Bayes, rely on this principle.