Fundamentals of GenAI cheat sheet
All 8 AIF-C01 flashcards for Fundamentals of GenAI on one page, grouped by exam objective. Print it or save it as a PDF for a last-minute read.
Explain the basic concepts of generative AI
- What is a token?
- The unit of text a model reads and writes as one piece of meaning. It can be a whole word, part of a word such as "-ed", or a punctuation mark.
- What is an embedding?
- A list of numbers (a vector) that captures the meaning of text or an image. Similar meanings get similar vectors, so you can search by meaning.
- What is chunking, and why do it?
- Splitting documents into smaller pieces before turning them into embeddings. Smaller pieces let a knowledge base retrieve just the relevant passage.
- Transformer vs diffusion model?
- Transformers weigh how each part of the input relates to the rest, which suits text. Diffusion models make images by adding noise to data, then removing it step by step.
- What is an AI agent?
- An application that uses a foundation model to work through a task in cycles. It interprets the input, decides what to do next and acts, for example by calling tools.
Understand the capabilities and limitations of GenAI for solving business problems
- Why can the same prompt get different answers?
- Foundation models are probabilistic: they pick likely next words rather than follow fixed rules. So identical inputs can produce different outputs (nondeterminism).
- Choosing a model: what matters besides quality?
- Cost, latency and must-have features such as image input or language support. Bigger isn't always better: a smaller, faster model can serve users better.
Describe AWS infrastructure and technologies for building GenAI applications
- Why use a managed service like Amazon Bedrock?
- AWS runs the infrastructure, so you call foundation models through an API. You build and scale without managing servers, with enterprise security built in.