AI glossary

A simple glossary of AI terms for business owners, management and teams who want to understand the technology, regulation and basic risks without unnecessary jargon.

AI system

A technology system that uses models or rules to create answers, recommendations, predictions or decisions from data.

AI tool

Software that uses artificial intelligence to help with writing, analysis, search, automation or data processing. Example: a tool that helps write a customer response.

AI agent

AI that can plan steps and perform tasks through tools, APIs or applications. Example: an agent that reads a request, checks data and prepares a draft answer.

Prompt

An instruction a user gives to an AI tool. Example: “Write a short customer response in a professional tone.”

AI model

A system trained on data that recognises patterns and generates responses. Example: a language model that writes text based on a user instruction.

Training data

Data used to train an AI model to recognise patterns.

Generative AI

AI that creates new content: text, images, code, summaries, presentations or draft responses.

Chatbot

A program that communicates with a user through conversation, most often through text.

Copilot

An AI assistant built into a work tool, for example for writing, spreadsheets, coding or presentations.

Hallucination

A situation where AI confidently provides incorrect or invented information.

Human oversight

A process where a person checks, approves or stops AI system outputs.

Risk category

A risk level that determines how strict the obligations are for a specific AI use.

High-risk AI

An AI system that can significantly affect people’s rights, safety or life opportunities.

Prohibited AI practices

AI uses that EU regulation prohibits because of unacceptable risk.

Deployer

An organisation that uses an AI system in its own operations, even if it did not develop it.

Provider

An organisation that develops or places an AI system on the market under its own name.

General-purpose AI model

A general-purpose AI model that can be used for many different tasks.

AI inventory

A list of AI tools and AI use cases used in a company.

AI usage policy

An internal document explaining what employees may and may not do with AI tools.

AI literacy

Basic understanding of the capabilities, limits and risks of AI tools.

Personal data

Data that can be linked to a specific person, such as a name, e-mail address or user identifier.

Sensitive data

Particularly risky data, such as health, biometric, political or religious data.

Biometric data

Data describing physical or behavioural characteristics of a person, such as face, voice or gaze.

Transparency

Clarity for users and employees about when AI is used and for what purpose.

AI audit

A structured review of AI use, risks, documentation, data, processes and responsibilities.

Automation

Using technology to perform repetitive tasks without constant manual work.

Workflow automation

Connecting steps in a business process so tasks are completed faster and more consistently.

Machine learning

A field of AI where systems learn patterns from data instead of every step being programmed manually.

Deep learning

A type of machine learning that uses multi-layered models for complex data.

Natural language processing

Technology that enables computers to understand and generate human language.

Large language model

An AI model trained to understand and create text, such as models behind modern chatbots.

Embedding

A numerical representation of text, image or data that helps a system find similarities.

RAG

An approach where AI searches relevant documents or a knowledge base before answering.

Fine-tuning

Additional training of a model on specific data so it performs better for a particular purpose.

API

A technical interface through which different systems connect and exchange data.

DPIA

A data protection impact assessment, often relevant when data processing can be risky.

Profiling

Automated analysis of personal data to assess characteristics, behaviour or interests of a person.

Automated decision-making

Making a decision by a system without meaningful human intervention.

Bias

An unwanted distortion of results caused by data, system design or the way a tool is used.

Model card

A short document describing an AI model, its purpose, limits and basic usage information.