Essential AI Terms

A foundational vocabulary linking data, algorithms, models, training, inference, and evaluation.

In this lesson

Vocabulary map

Data: examples, measurements, text, or images used by a system.
Algorithm: mathematical or logical steps for processing data or training.
Model: a structure and learned parameters representing patterns.
Training: adjusting model parameters using data.
Inference: running a trained model on new input.
Feature: a useful property used for learning or prediction.
Parameter: an internal value adjusted during training.
Dataset: an organized collection of examples.

Classification and regression

Classification chooses a category, such as spam or not spam. Regression predicts a number, such as price or arrival time. Generation creates new content based on patterns and context.

Accuracy is not the only metric

A reported accuracy of 99% may hide a serious failure when the important case is rare. The right metric depends on the cost of error: is accepting fraud worse than rejecting a valid transaction?

Quick check

What is the difference between an algorithm and a model?
Is predicting a house price classification or regression?
When does inference begin?