Xenith Guru
Glossary
The words you will meet, in plain English.
- AI
- Artificial intelligence: computer systems that do tasks we usually think need human intelligence, such as writing, recognising images or answering questions.
- Generative AI
- AI that produces new text, images, sound or video in response to what you ask.
- Large language model (LLM)
- A program trained on a vast amount of text to predict likely next words. Chat assistants are built on them.
- Chat assistant
- A tool you talk to in plain language, built on a large language model.
- Model
- The trained program itself, as distinct from the app or website you use it through.
- Provider
- The company or organisation that runs the model or the service.
- Prompt
- What you type or say to the assistant: your question or instruction.
- Context
- Everything the assistant can see while it answers: your prompt, the conversation so far, and any documents you have given it.
- Context window
- The limit on how much text an assistant can take into account at once. Very long conversations or documents may be partly dropped.
- Token
- A small piece of text, often part of a word, that models read and write in. Limits and prices are often counted in tokens.
- Training data
- The text, images or other material a model learned its patterns from.
- Training cut-off
- The date after which a model learned nothing new. It may not know about later events unless it is connected to search.
- Hallucination
- An answer that sounds right but is false or made up, such as an invented book or a wrong date.
- Source
- Where a claim comes from. A source only counts once you have opened it and found the claim in it.
- Bias
- A tilt in answers that comes from patterns in the training data or the way a system was built, and can treat some people or views unfairly.
- Local model
- A model that runs on your own device, so what you type does not leave it.
- Privacy notice
- The provider's statement of what it does with what you type and with your account.
- Agent
- An AI system that takes actions as well as answering, such as searching, filling in forms or running steps for you. Its actions need the same checking as its words.