What is diffusion model AI?

Diffusion Models are generative models, which means they generate data that resembles the data on which they were trained. Diffusion Models work by destroying training data by adding Gaussian noise to it over and over again, and then learning how to get the data back by reversing this process of adding noise.