AI glossary · Beginner · Language models
Unsupervised learning
Unsupervised learning is a machine learning approach in which the AI is given only data without answers attached and, on its own, groups similar items or finds hidden structure in it.
About this term
- Level
- Beginner · Core ideas for first-time readers
- Category
- Basics
- Fields
- Language models
- Added
- 2026-10-08
How it works
Unsupervised learning has no answers such as 'this is a cat.' The model finds patterns by looking at how similar data points are and which features appear together. A typical task is clustering, which groups similar data together. Dimensionality reduction, which simplifies data with too many features by keeping only the essentials, and anomaly detection, which finds values that stand apart from the rest of the data, are also commonly used.
Why it matters
Most of the world's data has no answers attached, and having people label it one by one is expensive. Unsupervised learning is valuable because it can use such data as is. It also helps uncover groupings or patterns people had not thought of. Approaches that create training signals from the data itself (self-supervised learning), such as the pretraining of today's LLMs, are connected to this line of work in that they also use large amounts of data without answers.
Things to know
Because there are no answers, it is hard to judge whether the results are 'right.' The same data can produce entirely different groupings depending on the method or settings. In the end, people have to check and interpret whether the groups found are actually meaningful. That is why, in practice, it is often used in an exploratory stage for understanding data or as a preparatory step for other kinds of learning.
Example
Online stores often analyze customers' purchase histories to divide shoppers into several groups. Even without predefining answers such as 'bargain hunter' or 'impulse buyer,' customers with similar purchase times, amounts and frequently bought items are grouped together automatically. Marketers look at these groups and design discounts or recommendations suited to each one.