Module 1 · Introduction to Generative AI and LLMs

OBJECTIVE: understand what generative AI is, how Large Language Models (LLMs) work in simple terms, which models exist, and what they can and cannot do — before we build with them. WHAT WE COVER: • What generative AI is: text, images, audio, code • How an LLM works: predicting the next token, pretraining, instruction tuning, human feedback • Tokens, context window and model size — the words you will see everywhere • The model landscape: closed models (GPT, Claude, Gemini) vs open models (Llama, Mistral, Qwen, Gemma), and small models • How to choose a model: quality, cost, speed, privacy, licence • Limits and risks: hallucination, knowledge cut-off, bias, privacy, responsible use • Three ways to use an LLM: chat app, API, running it locally AFTER THIS MODULE YOU CAN: • Explain how ChatGPT-like models work in plain words • Choose a suitable model for a business use case and justify it • Call an LLM from Python through an API and run one on your own laptop MINI EXERCISE: send the same 3 prompts to a cloud model (API) and a local model (Ollama) and compare answers, speed and cost.