No jargon, no background needed — just the handful of words that unlock most conversations about AI.

Agent — AI that can go and do tasks for you — click, search, take steps — not just answer in a chat.

AI (artificial intelligence) — A broad term for machines doing things that need intelligence when people do them. So broad it's almost meaningless on its own — always ask which kind.

Bias — Skew an AI inherits from its training data or design — worth asking about wherever it affects real people.

Chatbot — The app you actually talk to (ChatGPT, Claude, Gemini). A friendly wrapper around a model.

Generative AI — AI that creates new things — writing, pictures, audio, code — rather than just recognising or sorting. The kind behind the recent boom.

Hallucination — When AI makes something up but says it with total confidence. The reason to double-check anything that matters.

LLM (large language model) — The technology behind chatbots like ChatGPT — trained to predict the next word across enormous amounts of text.

Machine learning — Systems that learn patterns from examples instead of being programmed with fixed rules. The foundation under most of what we call AI.

Model — The trained “brain” itself, as distinct from the app wrapped around it.

Multimodal — AI that also handles images, audio, or video, not just text — it can “look” and “listen”, not only read.

Prompt — Whatever you type to ask the AI something. Clearer questions get better answers.

Training data — The text and images an AI learned from, frozen at a cut-off date — the source of both its knowledge and its blind spots.