Module 02
How AI Answers Are Built
An AI answer is a retrieval result with a writer on top. Learn every stage of that pipeline, so a change to it is something you can reason about rather than react to.
6 lessons · 12 videos · 8h 19m
- 02.01
Embeddings: Matching a Page That Shares No Words
Explain how text becomes a vector, why cosine similarity retrieves a relevant page containing none of the query's words, and what that model is blind to.
- 02.02
Grounding: An AI Answer Is a Search Result With a Writer on Top
Trace a generative answer back through grounding to the retrieved documents it was built from, and say which stage a missing citation actually failed at.
- 02.03
Query Fan-Out: One Question Becomes Twelve Searches
Describe how a single prompt is expanded into a set of concurrent related queries, and explain why that makes topic coverage matter more than any single keyword.
- 02.04
Rerankers: How a Passage Becomes a Citation
Explain what a reranker does that first-stage retrieval cannot, and describe the path from candidate document to the specific passage a model cites.
- 02.05
The Answer Engines, and Whose Index Each One Uses
Name what actually backs AI Overviews, AI Mode, ChatGPT Search, Copilot and Perplexity, and explain why being absent from one index removes you from more than one product.
- 02.06
Why the Same Prompt Answers Differently Twice
Explain the sources of non-determinism in a generated answer - sampling, personalisation, freshness, fan-out variance - and say what that does to any measurement built on prompts.