AI words in plain English.

Artificial Intelligence (AI)

Software that does things we used to think needed human intelligence — writing, answering questions, recognising images — by learning patterns from huge amounts of data.

Large Language Model (LLM)

The kind of AI behind ChatGPT and Claude. It learned from enormous amounts of text and works by predicting the most likely next words — which is how it writes and answers.

Prompt

What you type to an AI — your question, instruction or request. A clearer prompt gives a better answer.

Prompt engineering

The skill of writing prompts that get great results — giving the AI a role, context, examples and the exact format you want. Learn it here →

Hallucination

When an AI says something false but sounds confident. It happens because the model predicts plausible words, not verified facts — so always check anything important. How to catch it →

Token

A chunk of text — roughly a few characters or part of a word — that the AI reads and writes in. Usage limits and pricing are often counted in tokens.

Context window

How much text an AI can keep in mind at once — your conversation plus anything you paste. A bigger window means it can handle longer documents.

Generative AI

AI that creates new content — text, images, audio, video or code — rather than just sorting or labelling existing data.

Machine learning

The broader field where software learns patterns from examples instead of being programmed with fixed rules. AI models are built using it.

Model

The trained AI itself — the thing that takes your prompt and produces an answer. ChatGPT, Claude and Gemini are products built on models.

System prompt

Hidden background instructions that set an AI's role and rules before your conversation starts, shaping how it responds.

RAG (Retrieval-Augmented Generation)

A method where the AI looks up real information from your documents or a database before answering — so it leans on facts in front of it instead of memory.

AI agent

An AI that can take steps to finish a task — using tools, browsing or running actions — rather than only replying with text.

Multimodal

An AI that handles more than text — it can also understand or create images, audio or video.

Fine-tuning

Further training a model on specific examples so it gets better at a particular task or voice.

Training data

The huge collection of text, images or other content a model learned from. Its strengths and blind spots trace back to this data.

Parameters

The internal "dials" a model adjusts during training. More parameters can mean more capability, but also more cost to run.

Temperature

A setting that controls how predictable or creative an AI's output is — lower is safer and more consistent, higher is more varied.

Embeddings

A way of turning text into numbers that capture meaning, so software can find related content and "understand" similarity.

API

A way for other software to talk to an AI directly, so developers can build the model into their own apps and tools.

Inference

The moment an AI actually runs to produce an answer from your prompt — as opposed to the earlier training phase.

Open-source model

An AI model whose code and weights are freely available, so anyone can run, study or adapt it.

Chatbot

A program you talk to in plain language. Modern ones are powered by large language models, which is why they feel so capable.

AGI (Artificial General Intelligence)

A hypothetical future AI that could match humans across almost any task. Today's models are narrower and not there yet.

Memory / Projects

Features that let an AI remember your preferences or keep files and context across chats, so you stop re-explaining yourself.

Knowledge cutoff

The point in time after which a model wasn't trained on new information — so it may not know recent events unless it can search.

No term matches that — try another word. ✨