New to AI? You only need three labels to begin: the company makes the technology, the model is the engine, and the app is what you open. A method is a recipe developers use to build or teach AI.

Made byIan Goodfellow and fellow researchers
What kind of thing?AI building method
Can I use it directly?GAN is a method, not one company or one app. Developers build tools with it.
An open notebook with a speech bubble, paper boat and bright light bulb representing words and ideas.
A question, a picture or an idea can be a starting point. The model’s job depends on what it was built to do.

See the idea in three steps

Start withExamples of real faces
AI helps withCreator + checker
You getAn imaginary face

A simple illustration of the idea. Results depend on the model, training and app.

A familiar way to picture it

An everyday analogyImagine a trainee artist and an art checker. The artist makes pictures; the checker compares them with real examples.

An analogy helps you understand the main idea. It is not a description of every technical detail.

A simple GAN example

One part creates an imaginary face. The other checks whether it looks like a real face.

1

The creating part makes a new sample, such as a face.

2

The checking part tries to tell generated samples from real ones.

3

During training, feedback helps the creating part improve.

Does every image generator use a GAN?

No. GAN is one approach to building generative AI. Other image tools can use different approaches. The fact that an app makes images does not tell you which method it uses.

Where can I try it or learn more?

GAN is a method, not one company or one app. Developers build tools with it.

Original research ↗

A maker’s page or research page may explain the technology rather than provide a chat box. Available models, features and access can change.

A small activity for you

Describe the two jobs in your own words: who makes the sample, and who checks it? You do not need to write code to understand this idea.

You do not need to memorise model sizes or write a program to understand this lesson. First explain the idea in your own words. If you want to build an app later, the official resources are your next step.

Common beginner questions

Does every image generator use a GAN?

No. GAN is one approach to building generative AI. Other image tools can use different approaches. The fact that an app makes images does not tell you which method it uses.

Who made GAN?

Ian Goodfellow and fellow researchers. GAN is a method, not one company or one app. Developers build tools with it.

How can I start learning this topic?

Describe the two jobs in your own words: who makes the sample, and who checks it? You do not need to write code to understand this idea.

Keep learning

These related lessons help you compare the names without getting lost in technical details.

Convolutional Autoencoder Explained with a Simple Picture Example →What Is Gemini Omni? Google’s AI Video Model Explained →

See all 20 AI names and the seven-day learning plan →

Sources and review date

Reviewed 3 November 2026. Examples and analogies are written for beginners. Product facts are linked to the maker, original project or research below.