Generative AI

Not all AI creates. Some AI makes decisions or sorts things into groups: is this email spam or not, is this photo a cat or a dog. Generative AI is different because it produces new content. You give it a request and it writes, draws, or composes something in response.

The tools that have made AI famous, such as ChatGPT, Claude, and Gemini, are all generative. When you ask one to draft an email or explain a tricky idea, it is generating text word by word, shaping a reply that fits your request. Image tools do the same with pictures, building an image from your description.

For you, the practical takeaway is that generative AI is a creative partner, not a search engine. A search engine finds pages that already exist. Generative AI makes something new each time, which is powerful but also means two people can ask the same question and get slightly different answers.

Here is a small example that makes it concrete. Imagine you type “write a four-line birthday poem for my dad, who loves fishing and terrible puns” into a tool like ChatGPT. It composes a brand new poem on the spot, one that has never existed before and is shaped entirely around the details you gave. Ask again and you will get a different poem, because it is creating rather than looking one up.

Most text-based generative AI is built on a large language model, and the way it produces content sits within the wider world of artificial intelligence. Tools that can handle images and sound as well as text are described as multimodal.

If you want to practise telling the two kinds apart, our guide to which tasks are generative AI tasks has plenty of everyday examples and a quick test you can use.