Clustering
Also: cluster analysis
An unsupervised task with many algorithms that define "similar" differently: centroid-based (k-means), density-based (DBSCAN), hierarchical (dendrograms) and probabilistic (Gaussian mixtures). Results depend heavily on the distance metric and feature scaling.
Example
Points at 1, 2, 10 and 12 on a line naturally form two clusters, {1, 2} and {10, 12}, without any labels telling the algorithm so.