Why This Lesson Exists
Every field has jargon, but AI moves fast enough that the jargon changes monthly. This glossary covers the words you'll actually hear, not the ones that exist to impress people at conferences.
The Essential 15
AI (Artificial Intelligence) - Software that does things that used to need human intelligence. Broad umbrella term. Everything below lives under it.
Machine Learning - The way modern AI learns: showing a system lots of examples until it spots patterns, instead of programming rules by hand.
LLM (Large Language Model) - The engine behind ChatGPT, Claude, and friends. A model trained on enormous amounts of text to predict and generate language. "Large" refers to the size of the model.
Token - The piece of text a model processes at once. Roughly 4 characters, or about 3/4 of a word. "1,000 tokens" is roughly 750 words. You'll see this in pricing and limits.
Prompt - The text you type into an AI. Prompts are the new keyboard skill - how you phrase them changes everything you get back.
Prompt Engineering - The craft of writing prompts that get good results. Fancy name, simple activity: asking well.
Hallucination - When AI confidently makes things up. The single most important AI word to know. See the lesson on how AI works.
Context Window - How much text the model can see at once. Bigger window = it can handle longer documents and remember more of your conversation.
Fine-tuning - Taking a general model and training it a bit more on specific data to make it better at a niche task. Like sending a generalist employee to a specialist course.
RAG (Retrieval-Augmented Generation) - Giving the AI access to a searchable set of your documents so it can answer from your data instead of guessing. The tech behind "chat with your PDF".
Agent - An AI that doesn't just answer - it takes actions. Book the meeting, send the email, run the workflow. The current frontier of AI.
Embedding - Turning text into numbers so the computer can compare meaning. This is how "find similar documents" works.
Diffusion Model - The tech behind image and video generators. It learns by slowly adding noise to images, then learning to reverse the process. Yes, really.
Inference - The act of the model running and producing an answer. Training is the expensive part; inference is what happens every time you ask something.
AGI (Artificial General Intelligence) - The hypothetical AI that matches human intelligence across everything. We're not there. Anyone who says otherwise is selling something.
Terms You Can Politely Ignore
Most of the alphabet soup (RLHF, MoE, LoRA, transformers) is implementation detail. Useful if you're building AI, unnecessary if you're using it. Don't let acronyms gatekeep you - the tools work the same whether or not you know what's inside.
Try This Today
Pick any AI chatbot and ask it: "Quiz me on these 15 AI terms. Give me one at a time, and explain the ones I get wrong." You'll have them memorised by lunch.
The One-Line Summary
You need about 15 words to talk about AI like you belong in the room - and now you have them.