Autoencoders are neural networks designed to learn efficient encodings of input data. They compress input into a lower-dimensional latent representation and then reconstruct the original data from this encoding. Variants like denoising autoencoders and variational autoencoders are used for noise removal and generative modeling.