Within recent years Machine Learning have been successfully used to solve a wide range of problems in physics. From training Neural Networks to find many-body particle wave functions to controlling large experiments such as creation of Bose-Einstein condensates or even a fusion reactor.  In this colloquium the subfield Reinforcement Learning is presented and used to solve a quantum state preparation problem.  Reinforcement Learning is computer learning based solely on gained experience. The computer is able to learn without having any prior knowledge of the problem at hand. This makes it suitable for solving quantum control problems where the underlying dynamics may not be intuitive or the optimal solution difficult to find. The algorithm used, a version of Watkins’ Q-learning algorithm, is tested on a two-level quantum system whose dynamics is governed by a bang-bang Hamiltonian. The algorithm successfully outperforms other numerical methods.