AI Glossary

A

AI Agent

An AI system that can plan and carry out multi-step tasks by itself, using tools such as web search, files, apps or code, rather than only replying to one message at a time.

AI Bias

When an AI system gives unfair or one-sided results for certain groups or situations, usually because of patterns in the data it learned from or in how it was built.

Artificial Intelligence (AI)

Computer systems that perform tasks that normally need human intelligence, such as understanding language, recognising images, making predictions and generating text or pictures.

C

Chatbot (AI Assistant)

A program you talk to in a chat, by typing or speaking. Modern AI chatbots use large language models to understand free-form questions and write natural replies.

Context Window

The maximum amount of text, measured in tokens, that an AI model can consider at one time, including your messages, uploaded documents and its own replies in the conversation.

D

Deepfake

An image, video or audio clip created or altered with AI to make it look or sound like a real person said or did something they never did.

F

Few-Shot Prompting

Including one or more examples of the output you want in your prompt, so the AI copies their style, length and format. One example is "one-shot", a few examples are "few-shot".

Fine-Tuning

Further training an existing AI model on a smaller, specific set of examples so it performs better at a particular task, style or domain.

G

Generative AI

AI that creates new content, such as text, images, audio, video or code, based on a request (prompt), instead of only analysing or classifying existing data.

H

Hallucination

When an AI tool gives information that sounds confident and plausible but is false or made up, such as a fake source, a wrong number or a rule that does not exist.

K

Knowledge Cut-off

The date up to which an AI model's training data goes. The model does not know about events, rule changes or new products after that date unless it can search the web or you give it the information.

L

Large Language Model (LLM)

An AI model trained on huge amounts of text that can understand and generate human-like language. LLMs power chat assistants like ChatGPT, Gemini, Claude and Copilot.

M

Machine Learning (ML)

A way of building AI where the computer learns patterns from examples (data) instead of being programmed with fixed rules for every situation.

Multimodal AI

AI that can work with more than one type of input or output, such as text, images, audio and video, for example reading a photo of a bill and answering questions about it.

P

Prompt

The instruction or question you give to an AI tool. A clear prompt with context, a specific task and the format you want gets much better answers than a short, vague one.

Prompt Chaining

Breaking a big task into a sequence of smaller prompts, checking each result before moving to the next step, instead of asking for everything in one giant prompt.

Prompt Engineering

The skill of writing and refining prompts so an AI tool gives accurate, useful and consistent results, using techniques like giving context, examples, structure and step-by-step instructions.

R

Retrieval-Augmented Generation (RAG)

A technique where an AI system first finds relevant information from a set of documents or a database, then uses it to write its answer, so responses are based on your actual content rather than only on what the model learned in training.

T

Token

A small piece of text, often part of a word, that language models read and write. AI tools measure input length, output length, limits and pricing in tokens.

Training Data

The examples an AI model learns from, such as text, images or records. The quality, variety and age of the training data shape what the model knows and where it makes mistakes.