A free guide by New Money School
Save This
Every one of these gets thrown around like you already know it. Here they are in plain English, each with a real example, so they finally stick.
Everyone in AI talks like you already know these words. You don't have to fake it anymore.
Here are the 10 terms that get thrown around the most, each one explained in plain English with a real example so it actually sticks. Read it once, save it, and come back whenever a word trips you up.
Term 1LLM (Large Language Model)
This is the brain behind the chatbot. It read a huge chunk of the internet and learned to do one thing really well: predict the next word. When you talk to Claude or ChatGPT, you're talking to an LLM.
Why it matters: it's basically the world's most well-read autocomplete. Once that clicks, a lot of what AI does stops feeling like magic and starts making sense.
Term 2Prompt
A prompt is just what you type to the AI. It's the instructions you give it. The trick is to be specific, because a vague question gets a vague answer. "Write something about marketing" gets you generic junk. "Write a 3-sentence Instagram caption for a life coach targeting burnt-out 9-to-5 workers, casual tone, end with a question" gets you something you can actually use.
Why it matters: the more detail you give, the better the answer. Most bad AI results are really just vague prompts.
Term 3Tokens
Tokens are how AI reads. It doesn't see whole words, it sees chunks. The word "basketball" might split into "basket" and "ball." All the limits and the cost of using AI are measured in tokens, not words.
Here's the rough math: 1 token is about three-quarters of a word, so 100 tokens is about 75 words, and 1 word is about 1.3 tokens. A typical page of text (around 375 words) is roughly 500 tokens. The exact counts vary a little by model. Why it matters: when you hear about token limits or per-token pricing, this is what they mean.
Term 4Context Window
The context window is the AI's short-term memory for one conversation. It's how much it can hold in its head at once. Go past it and it starts forgetting the beginning of the chat, which is why a really long conversation slowly gets worse.
The good news is that modern models hold a lot now. The newest Claude models, like Sonnet 5, can hold up to about 1 million tokens at once, which is roughly 750,000 words. Why it matters: if a long chat starts feeling forgetful, you've filled the window. Start a fresh chat and it's sharp again.
Term 5Hallucination
A hallucination is when AI makes something up but says it like it's 100% fact. It's not lying, it's guessing and sounding confident while it does. This is why you always double-check names, dates, and numbers.
Here's a real example. The Eiffel Tower was completed in 1889, designed and built by Gustave Eiffel's company (engineers Maurice Koechlin and Emile Nouguier, with architect Stephen Sauvestre) for the 1889 World's Fair marking 100 years since the French Revolution. If an AI confidently gave you a different year or a made-up engineer, that's a hallucination. Why it matters: AI is a brilliant first draft, not a final source. Verify anything that counts.
Term 6Machine Learning
Machine learning is how AI learns from examples instead of from rules someone typed in. Show it a ton of data, it spots the patterns, and it gets better on its own. Feed it thousands of emails labeled spam or not-spam and it learns the signals, like all-caps subject lines and weird links. The more it sees, the more its accuracy climbs.
Why it matters: nobody wrote those rules. The AI figured them out from the examples. That's the whole idea behind machine learning.
Term 7MCP / Connectors
Connectors, sometimes called MCP, are the plugs that let AI reach into your other tools like your email, calendar, files, and Slack. Instead of you copying and pasting everything in, the AI goes and gets it itself.
Why it matters: this is the part that makes AI feel like magic. Once it can see your real inbox and calendar, it stops being a chatbot and starts being an assistant.
Term 8Generative AI
Generative AI is the catch-all term for AI that makes brand-new stuff: text, images, video, and music. If it creates something instead of just sorting or analyzing what's already there, it's generative.
Why it matters: it's the umbrella word for most of the AI you hear about. Every chatbot and image tool you've tried lives under it.
Term 9AGI (Artificial General Intelligence)
AGI, or artificial general intelligence, is the hypothetical point where AI can do basically any task a human can, including things it was never trained on. We are not there yet.
Why it matters: when people argue online about AI "taking over," this is usually what they mean. It's a real debate, but it's about the future, not the tool you're using today.
Term 10Vibe Coding
Vibe coding means building software by just telling the AI what you want in plain English, instead of knowing how to code. You describe it, the AI writes it, and you tweak it from there.
Why it matters: it's why total beginners are suddenly shipping real apps. You don't need a computer science degree, you need a clear idea and the words to describe it.
Bonus5 More To Learn Next
You don't need these on day one, but they'll come up soon. Here's the one-line version of each so you're ready.
- Agent: an AI that doesn't just answer, it takes steps and gets a whole task done for you, start to finish.
- RAG: a way to let AI pull answers from your own documents, so it responds with your real info instead of just what it was trained on.
- Fine-tuning: extra training that teaches a model your specific style or task so it fits you better.
- Multimodal: AI that handles more than text, like images, audio, and video, all in one place.
- Open-source model: an AI model anyone can download, run, and change for free, instead of one locked behind a company.
The Real Win
Once these click, nothing in AI sounds scary again. You stop nodding along to words you don't know and start actually following what's going on. That's the whole game. Save this, come back to it, and watch how fast the jargon stops being jargon.