Reinforcement Learning (RL)
Also known as: RL
A way of training AI in which a system learns from the results of its own actions rather than from being shown correct answers. Richard S. Sutton and Andrew G. Barto, in their standard textbook, describe reinforcement learning as "learning what to do", mapping situations to actions "so as to maximize a numerical reward signal". The learner "is not told which actions to take, but instead must discover which actions yield the most reward by trying them", and the authors name "trial-and-error search and delayed reward" as its two most important distinguishing features. For businesses, the useful question about a model trained this way is what it was rewarded for, since that is what it learned to pursue; see also reward hacking.