AI that creates new things: text, images, audio or video.
Generative AI doesn't just sort or search existing things; it makes new text, images, audio and video. It learns patterns from huge numbers of examples, then follows those patterns to produce something new, piece by piece.
Text usually comes from large language models; images often come from diffusion models, which start from noise and clean it up step by step. Research such as GANs in 2014 laid groundwork, and in 2022 image generators and ChatGPT arrived one after another and took it mainstream. It also brings problems: false content, copyright questions and deepfakes. Check what it makes, and say when you used it.
Think of it like this
Picture a friend who has heard so much music they can hum a brand-new tune "in a summer-hit style". Generative AI likewise builds new things out of patterns it learned. It can sound convincing without always being right.
For example
Type "a whale flying over the sea at sunset" and get a newly drawn picture of that scene: that's generative AI.
Records about this
Source-checked moments from AI history.
The conversational UI opened prompting to non-experts, turning a skill of a few practitioners into a mass phenomenon
Made a high-quality generative model that runs on consumer GPUs open-weight, sparking both the LoRA and ControlNet ecosystem and the copyright debate
Proposed an architecture that uses the competition between a generator and a discriminator as the learning signal, opening up one pillar of generative modeling