📚 AI Fundamentals

Status: 🔜 Coming soon
Goal: Build a solid foundation of AI concepts — the terminology, how it all fits together, and what it means in practice

Why This Matters

As a Copilot Solution Engineer, customers ask "what's an LLM?", "how is this different from ML?", "what does grounding mean?" — you need crisp, confident answers.

Planned Topics

Topic	Description
The AI family tree	AI → ML → Deep Learning → GenAI → LLMs (nested dolls diagram)
What is Machine Learning?	Learning from data without explicit programming — 3 types: supervised, unsupervised, reinforcement
What is Deep Learning?	Neural networks with many layers — why they're powerful for images, text, speech
What is a Large Language Model?	How LLMs work: training, tokens, context windows, parameters, temperature
What is Generative AI?	The category that creates new content (text, images, code, video)
Foundation Models	The "base" models everything is built on (GPT, Claude, Gemini, Llama)
Tokens & context windows	The currency of AI — why 128K vs 1M matters
Hallucinations & grounding	Why AI makes stuff up and how RAG/grounding fixes it
RAG explained	Retrieval Augmented Generation — how Copilot gets your real data into answers
Fine-tuning vs prompting	When you customise the model vs when you just ask differently
AI safety & responsible AI	Bias, ethics, Microsoft's responsible AI principles
The AI glossary	Quick-reference table: 50+ terms defined in simple English

☕ Café Analogy

An LLM is like a chef who's read every cookbook ever written — they know recipes from memory but have never tasted the food. RAG is like giving them today's menu and ingredient list so they cook what's actually available.

The AI Family Tree

🤖 Artificial Intelligence (AI)
└── 📊 Machine Learning (ML)
    └── 🧠 Deep Learning
        ├── 🗣️ Large Language Models (LLMs)
        ├── 🔬 Small Language Models (SLMs)
        ├── 🎨 Image Models (DALL-E, Midjourney)
        └── 🎬 Video Models (Sora)
    = All of this is "Generative AI" when it creates new content