// ai glossary
AI words, explained simply
Every important AI term in plain English. Bookmark this page and come back whenever you meet a new word.
- Agent (AI agent)
- An AI system that can plan steps and use tools (like search, email or code) to complete a goal, not just answer one message.
- Example: An agent that reads new support emails, looks up the order and drafts a reply.
- Algorithm
- A set of steps a computer follows to solve a problem.
- API
- A way for one program to talk to another. Developers use AI APIs to add AI features to their own apps.
- Artificial intelligence (AI)
- Computers doing tasks that normally need human thinking, such as understanding language or recognising images.
- Bias
- When an AI's answers unfairly favour or disfavour certain people or ideas, often because of patterns in its training data.
- Chatbot
- A program you talk to in normal language. Modern chatbots like ChatGPT, Claude and Gemini are powered by large language models.
- Context window
- How much text (measured in tokens) a model can consider at once, including your messages and its replies.
- Embedding
- A list of numbers that represents the meaning of a piece of text, so a computer can compare meanings. Used in search and RAG.
- Few-shot prompting
- Giving the AI a few examples of the output you want inside your prompt.
- Example: Two sample taglines, then: 'Write 3 more in this style.'
- Fine-tuning
- Training an existing model a bit more on your own examples so it gets better at a specific task or style.
- Generative AI
- AI that creates new content, such as text, images, audio, video or code.
- Hallucination
- When an AI states something false as if it were true, such as a made-up fact, number or source.
- Inference
- The moment a trained model is used to produce an answer. Each chat reply is an inference.
- Large language model (LLM)
- An AI model trained on a very large amount of text so it can understand and generate language.
- Machine learning
- Building software that learns patterns from data instead of following only hand-written rules.
- Model
- The trained AI 'brain' that produces outputs. Companies release different models with different strengths and prices.
- Multimodal
- An AI that can work with more than one type of input or output, such as text, images and audio.
- Neural network
- A type of machine-learning model made of many connected layers of simple maths units, loosely inspired by the brain.
- Open-source model
- A model whose weights are publicly released so anyone can run it on their own hardware, within its licence.
- Parameters
- The internal numbers a model learns during training. More parameters usually means a bigger, more capable (and more expensive) model.
- Prompt
- The instructions or question you give an AI.
- Prompt engineering
- Designing prompts on purpose so the AI gives reliable, useful results.
- RAG (retrieval-augmented generation)
- A method where the AI first searches your documents for relevant information, then uses it to write the answer. Reduces made-up facts.
- Example: A company chatbot that answers from the official policy documents.
- System prompt
- Hidden instructions a developer gives a model to set its role, rules and style before the user starts chatting.
- Temperature
- A setting that controls how random or creative a model's output is. Low = more predictable, high = more varied.
- Token
- A small piece of text (often part of a word) that models read and write. Usage and limits are measured in tokens.
- Training data
- The examples a model learned from. Its knowledge and biases come from this data.
- Vector database
- A database that stores embeddings so you can quickly find text with similar meaning. Common in RAG apps.