Machine learning is the study of computer programs that get better at a task by learning from data instead of following hand-written rules. The term traces to Arthur Samuel, whose 1959 IBM Journal paper “Some Studies in Machine Learning Using the Game of Checkers” described a checkers-playing program that improved its play through repeated games. Samuel is widely credited with popularizing the phrase “machine learning” through this work.
In modern practice, a machine learning system is given examples (training data), a way to measure how well it is doing (a loss function), and a method for adjusting itself to do better (such as gradient descent). The “Machine Learning Basics” chapter of Goodfellow, Bengio, and Courville’s “Deep Learning” frames a learning algorithm as one whose performance on a task improves with experience, and organizes the field into supervised, unsupervised, and other learning settings.
The key shift is that the programmer specifies the goal and the data, not the step-by-step solution. The program discovers the rules itself by finding patterns that generalize from past examples to new ones.
Why business readers should care: Machine learning lets software handle tasks too messy or varied to spell out by hand, like detecting fraud, ranking search results, or forecasting demand. Understanding that these systems learn from data, and are only as good as that data, is the foundation for using them responsibly.