Generative AI
Explore the technologies behind AI-generated text, images, code and other content, and learn where generation belongs in a product.
This series covers what generative models can produce, how they work, how they are steered and evaluated, and the governance questions they raise.
It assumes familiarity with LLM Basics. Each topic cites the foundational papers and guidance it draws on.
What You’ll Learn
Discriminative models learn to separate categories, such as spam versus not spam. Generative models learn the distribution of the data itself, so they can produce new samples: text, images, audio or code. Many work by sampling from a learned latent space.
A discriminative model answers “which one is this?”. A generative model answers “make me a new one”.
Generative adversarial networks, introduced in 2014, trained a generator against a discriminator and produced some of the first convincing synthetic images.