In plain English
Unsupervised learning is a machine learning approach where the model finds patterns in data without labelled examples to guide it. With no answer key, it instead surfaces structure on its own, which makes it useful for clustering similar items, detecting anomalies, and simplifying complex data. It is valuable precisely when labelled data is scarce or expensive.
A worked example
A retailer feeds purchase histories into an unsupervised model, which groups customers into natural segments without anyone defining the groups in advance.
Common confusion
Unsupervised does not mean unattended or fully automatic. People still choose the data, the method, and how to interpret the patterns it finds.