A Glossary of AI Terms for Non-Technical Teams

AI conversations move quickly and the vocabulary is unfamiliar. This glossary explains the terms you are most likely to hear, in plain English and without maths.

Core concepts

• Artificial intelligence (AI): software that performs tasks we associate with human intelligence, such as understanding language, recognising images or making predictions.
• Machine learning: a way of building AI by letting software learn patterns from examples instead of following hand-written rules.
• Generative AI: AI that creates new content: text, images, audio, code or video.
• Large language model (LLM): a generative model trained on large amounts of text to predict and produce language. Chat assistants are built on LLMs.
• Model: the trained system itself, the thing that takes an input and produces an output.

Using a model

• Prompt: the instruction or question you give the model. See our prompt-writing framework.
• Token: a chunk of text, often part of a word, that models read and write. Pricing and limits are usually counted in tokens.
• Context window: how much text the model can consider at once, including your prompt, any documents and its own reply.
• Temperature: a setting that controls how varied or predictable the output is.
• Multimodal: able to work with more than one type of input or output, such as text and images.

Making models more useful

• Embedding: a list of numbers representing the meaning of a piece of text, used to find similar passages.
• Retrieval-augmented generation (RAG): giving a model relevant documents at question time so it answers from them. Explained in our RAG guide.
• Fine-tuning: further training a model on specific examples to change its behaviour or style.
• Agent: an AI system that can take actions, such as searching, using software tools or completing multi-step tasks, rather than only replying with text.
• API: the interface developers use to connect a model to other software.

Risks and governance

• Hallucination: confident output that is false.
• Bias: systematic unfairness in outputs, often inherited from training data.
• Training data: the material a model learned from.
• Human in the loop: a process where a person reviews or approves AI output before it is used.
• AI governance: the policies, roles and checks an organisation uses to manage AI risk. The NIST AI Risk Management Framework is a widely used free reference.

Keep this list handy for vendor meetings: asking "How does this handle hallucinations?" or "Is our data used for training?" is often the fastest way to understand what a product really does.

Cover photo by majed swan on Unsplash.