TL;DR - You don't need deep theory, but you do need the vocabulary. Knowing ~15 core terms lets you follow AI conversations and make decisions instead of nodding along.
Why it matters
Workers who can't speak the language get left out of decisions about AI at work. These terms also unlock the rest of this journey - you'll see them in every later level.
The essential terms
- Model - the AI system (GPT, Claude). Prompt - your input. Token - a chunk of text it reads/writes (you're billed by these).
- Context window - how much text it can consider at once. Hallucination - confident wrong output.
- Zero-shot / few-shot - no examples vs a few examples in your prompt. Temperature - the creativity dial.
- System prompt - standing instructions that shape every reply. RAG - feeding the model your own documents.
- Agent - AI that uses tools to take actions. Multimodal - handles text + images/audio. Fine-tuning - extra training on specific data. MCP - a standard way to connect AI to tools.
Worked example - use them in a sentence
"Let's give it a few-shot example and lower the temperature so the format stays consistent, and put the long input near the end because of the context window."
If that sentence makes sense, you can hold your own in an AI conversation.
Steal this - the flashcard set
model · prompt · token · context window · hallucination
zero-shot · few-shot · temperature · system prompt
RAG · agent · multimodal · fine-tuning · MCP
Goal: use 10 of these correctly in conversation.Common mistakes
- Nodding along instead of asking "what does that mean?"
- Confusing fine-tuning with RAG - fine-tuning re-trains; RAG just feeds documents at answer time.
- Thinking "agent" = chatbot - an agent acts using tools.
Good to know
You'll go deep on most of these later: prompting terms in Level 2, agents and automation in Level 4, RAG and MCP in Level 5. Here, just get comfortable saying them - fluency in the words is the on-ramp to fluency in the skills.