Tianhua Chen has released a 195-page preprint on arXiv, "The Little Book of Generative AI Foundations: An Intuitive Mathematical Primer" (arXiv:2605.29713 [cs.LG]), first submitted May 28, 2026, with a revised version appearing September 8, 2026, that added visual explanations throughout. The book traces the mathematical lineage connecting PCA, probabilistic PCA, variational autoencoders, diffusion models, normalizing flows, autoregressive factorisations, GANs, Wasserstein GANs, and energy-based models, aiming to give researchers a unified foundation rather than isolated architectures. Rather than surveying every recent architecture or implementation detail, Chen develops the mathematical relationships linking these families, with the stated aim of making the structure of generative modelling more accessible while preserving the derivation-level substance needed to understand how these models are built and related. The target audience is mathematically curious researchers, practitioners, and students who want a foundation-building primer rather than a survey of papers. According to the article, diffusion models dominate image generation, GANs shaped early high-fidelity synthesis, variational autoencoders introduced the latent space idea later architectures built on, normalizing flows offered tractable likelihood estimation, and energy-based models connected statistical mechanics to learned distributions, with each family emerging from its own theoretical lineage that the field often treated as separate domains; Chen's book attempts to show how those lineages connect.