Diffusion model

A diffusion model is a type of generative AI ✕ Generative AI creates new content —text, images, audio or code— from a prompt, instead of only analyzing existing data. More in the glossary → model used above all to create images, though also audio or video. It works in a curious way: during training it learns to reverse a process that keeps adding "noise" (random speckling) to an image until it is completely destroyed. Once trained, it does exactly the opposite: it starts from pure random noise and gradually cleans it up, over many steps, until a sharp, coherent image emerges from that chaos. When combined with text, a written prompt ✕ A prompt is the instruction or question you give an AI to get a response. The clearer and more specific it is, the better the result tends to be. More in the glossary → guides the process so the result matches what we asked for. It is the technology behind well-known image generators such as Stable Diffusion, DALL·E and Midjourney. Diffusion models have largely displaced earlier techniques thanks to the quality and realism of their output, and they are a central piece of the recent boom in generative AI.