AI glossary¶

Short definitions of the terms used in the AI Gazette. The tips link here rather than defining them again.

Agent¶

A program that chains several actions and tools to reach a goal, whereas an assistant simply answers a question.

Anonymization¶

Replacing what identifies a person, a customer or a project (names, amounts, places) with neutral elements before sending a text to an AI. See the tip Data security and privacy.

Context window¶

The amount of text a model can take into account at once: your current conversation, your attachments and its own answers. Beyond that, or when it gets very long, the beginning of the conversation is taken into account less well.

Hallucination¶

False information presented confidently by an AI: a figure, a quote, a reference or a source that does not exist. See the tip Hallucinations: why AI makes things up.

LLM¶

Large Language Model: a program trained on very large amounts of text to produce the most likely continuation of a message. See the tip What is an LLM?.

Model¶

A specific version of an AI assistant, with its level of power, speed and cost. The colored dots in the interface show the level: 🟢 economy, 🟡 intermediate, 🔴 premium. See the tip Choosing your AI model.

Prompt¶

A message or instruction sent to the AI. Its quality determines the quality of the answer. See the tip Structuring your first message.

Token¶

A fragment of text, a word or a piece of a word, that the model handles. Text length and usage cost are often measured in tokens.