Machine Learning, a subset of artificial intelligence, is a family of computer algorithms that enable a model* to learn conditional probabilities of outcomes given inputs (the task) by observing data (the experience) and minimising some error measurement (the metric). ML is not the only way to approach this task; but, it has long been settled that ML produces models that work best in examples not seen in training.
*In Computer Science, a model is said to "learn" to perform a task, given a performance measure, if it's performance at the task improves with experience.