Glossary
Machine learning is the practice of building systems that derive their rules from data rather than having those rules written by a programmer. The system learns a pattern from examples and applies it to new cases.
Classifying an email, forecasting demand, scoring a lead and detecting an anomaly are all machine learning, and most of them use far simpler and cheaper methods than a language model.
A model learns from examples, so it needs enough of them, labelled correctly, representing the situations it will meet. Where a business lacks that history, machine learning is not yet the answer regardless of how well suited the problem sounds.
A model trained on past decisions reproduces the patterns in those decisions, including the ones nobody intended to encode, which is why the question of what the training data represents matters more than the choice of algorithm.
Machine learning is a subset of artificial intelligence, and generative AI is a subset of machine learning. Many business problems described as AI are better solved with simpler, cheaper machine learning methods.
It depends on the problem, but the harder constraint is usually quality rather than quantity: correctly labelled examples that represent the situations the system will actually meet.
Machine learning is the practice of building systems that derive their rules from data rather than having those rules written by a programmer. The system learns a pattern from examples and applies it to new cases.
