In this introduction, we lay out the conceptual and structural foundations of the book. We define machine learning for readers unfamiliar with the field, covering its main types (supervised, unsupervised, reinforcement, and generative learning) and the key stages of the ML pipeline: datasets, features, model training, testing, and output. We then examine what "human-centered" means in computing, tracing how the concept has expanded beyond usability to encompass well-being, social impact, and the needs of communities beyond individual users. We define human-centered ML as the application of a human-centered mindset to the design, use, and evaluation of ML, and then map out the book’s four parts: ML in the arts and humanities; the methods and practices of human-centered ML; humans in the ML process; and ML in its broader social context. It closes with perspectives from the book’s interviewees on what human-centered ML means to each of them: definitions that range from keeping humans in the loop throughout the entire pipeline to building ML from within affected communities to a critique of the "human-centered" framing as too narrow for a world in environmental crisis.