Learning path · Foundations · 01
What Is Generative AI
AI systems that create new text, code, images, or audio by learning statistical patterns from large datasets instead of executing only hand-written rules.
Why it matters
- Tells generative systems apart from classifiers and rule engines.
- Fluency is cheap. Grounding and reliability take design.
- Later pages are mostly about making generation usable at work.
Key ideas
- New artifacts, not lookups
- Foundation models
- Fluency is cheap
Generative models write text, code, images, or audio that look like their training data. That is different from looking up your policy PDF. The weights hold a fuzzy prior; they do not hold last week's tickets. Put retrieval, tools, and checks around the model if the answer has to match a system of record. Before you scale traffic, write down what must come from outside the model, and how you will see failures.
Updated 2026-08-09 · Full learning path