What is generative AI?

An AI system can classify a photo, predict demand, or recommend a song. Generative AI describes a different kind of task: producing a new output, such as a paragraph, image, audio clip, or computer-code suggestion. The output is generated from patterns learned during training, often guided by an input called a prompt.

Generation is a task, not one single model

Generative AI is an umbrella term for multiple approaches and types of models. A large language model generates text; other model families can generate images, audio, or other data. Some systems can work across more than one kind of input or output. The common feature is that they generate content rather than only assigning a label to an input.

What the prompt does

A prompt gives the system context and a request. For example, a student might ask for a plain-language explanation of a programming term or request a first draft of a study outline. The system uses the prompt, its learned patterns, and any available context to produce an output. Clearer context can help, but a better prompt cannot guarantee a correct answer.

Different generative systems work differently internally. It is more accurate to understand generative AI as a family of capabilities than to assume that every tool uses the same model or generation process.

How it relates to machine learning

Many generative AI systems are built with machine-learning methods. Machine learning is the broader set of techniques that learn patterns from data; generative AI describes systems that use learned patterns to create outputs. That is why understanding basic ideas such as training data, models, and inference makes generative AI easier to reason about.

What to remember