Generative AI
How models that create text, images, sound and other content work.
The chapter at a glance
PromptContext or retrieved sources
Generative model
Text • image • sound
A person checks
From “Context or retrieved sources” it loops back to “Generative model”.
In this chapter
- 1What is generative AI?Generative AI creates new outputs, such as text, images, speech, code or video.
- 2What is a large language model (LLM)?An LLM is a model trained on huge amounts of text or similar data in order to work with language.
- 3The transformer: how does a model connect words?The transformer is a highly influential neural-network design for processing sequences.
- 4What is a prompt, and how do you ask a good question?A prompt is the instruction or input we send to a model.
- 5Fine-tuning and human feedbackA pre-trained model can be adapted to particular tasks, languages or ways of answering.
- 6What is RAG, and why does AI read documents?RAG means the system looks up relevant information before answering and adds it to the context.
- 7Open or closed model: what do you actually get?“Open” can mean open code, downloadable weights, known training data or free use — and these are not the same thing.
- 8Why does AI “hallucinate”?“Hallucination” is the common name for an answer that sounds convincing but contains made-up or false claims.
- 9Smaller models and distillationProgress in AI is not always about making models bigger. A lot of research looks for more efficient solutions.
- 10What are world models?They try to represent how an environment changes and what consequences particular actions have.