1. Home
  2. Glossary
  3. Machine learning (ML)
AI glossary · Foundations

Machine learning (ML)

Machine learning (ML): Machine learning is the branch of AI in which software learns patterns from examples instead of following rules a programmer wrote by hand. Nearly every modern AI tool, including chat assistants, is built on it.

Traditional software works from explicit instructions: if the invoice total is over $10,000, route it to a director. Machine learning flips that. You show the system thousands of past invoices labeled 'approved' or 'flagged,' and it works out the patterns that separate them. The rules are learned, not written.

There are three broad flavors. Supervised learning trains on labeled examples (this email is spam, this one is not). Unsupervised learning finds structure in unlabeled data, like grouping customers by buying behavior. Reinforcement learning improves by trial and reward, the way a model learns which chess moves lead to wins.

The catch is that a model is only as good as its data. If your historical hiring data reflects past bias, a model trained on it will learn that bias too. And a model trained on last year's patterns can quietly degrade when the world changes. ML systems need ongoing monitoring, not a one-time setup.

Example at work

A logistics company wants to predict which shipments will arrive late. Instead of writing rules, an analyst feeds two years of shipment records (carrier, lane, weight, weather, day of week, actual arrival) into a machine learning model. The model learns which combinations tend to run late and scores each new shipment so dispatchers can intervene early.

Why it matters

Every AI feature you will be sold at work, from demand forecasting to the chat assistant in your CRM, is a machine learning system underneath. Understanding that it learned from examples tells you the right questions to ask a vendor: what data was it trained on, how is it monitored, and what happens when it is wrong?

Related terms