A model architecture where two neural networks compete: a generator creates fake data, and a discriminator tries to tell real from fake. Through this adversarial game, the generator gets better at creating realistic outputs. Dominated image generation from 2014 to ~2022.
Why it matters
GANs pioneered realistic AI image generation and are still used in some real-time applications. But diffusion models have largely replaced them for quality-critical work because GANs are harder to train and less diverse in their outputs.