Machine Learning
Etymology
The phrase was coined by IBM researcher Arthur Samuel in 1959, who defined it as giving computers the ability to learn without being explicitly programmed. "Learning" is a loose analogy — the system adjusts internal numbers to reduce its errors on examples, not to understand. The field languished for decades before cheap computing and vast datasets made it the dominant approach in the 2010s.
noun phrase · /muh-SHEEN LUR-ning/
Building systems that improve at a task by finding patterns in data rather than following hand-written rules.
What it names
The shift behind most modern AI: programs that are trained on examples instead of being explicitly told what to do.
Example
The spam filter uses machine learning, so it adapts as junk mail changes its tricks.
Related idea
Machine learning is a subset of artificial intelligence: AI is the broad goal of machines doing intelligent-seeming things, while machine learning is the specific, now-dominant method of getting there by learning from data.