AI TERMS FOR INSTITUTIONS

Short plain-language explanations of five terms that come up when institutions talk about AI, each with a link to the primary source. They are our paraphrases, not legal text: check the source before you rely on it.

AI system

A machine-based system that, from the inputs it receives, works out how to produce outputs such as predictions, content, recommendations or decisions, with varying levels of autonomy. The EU AI Act (Article 3) and the OECD (2023 update) rest on this core idea.SOURCES: EU AI Act, Article 3 ↗ · OECD · AI system definition (2023) ↗

Human oversight

People stay responsible for decisions made with AI and can step in. UNESCO’s Recommendation on the Ethics of AI, adopted by its 193 Member States in November 2021, lists human oversight among its ten principles.SOURCES: UNESCO · Recommendation on the Ethics of AI ↗

Transparency

Openness about when and how AI is used. UNESCO lists transparency among its ten principles. On this site it means marking AI-assisted texts and linking sources for facts.SOURCES: UNESCO · Recommendation on the Ethics of AI ↗

AI risk management

A routine for finding and reducing the risks AI poses to people, organisations and society. The US NIST framework, released on 26 January 2023 for voluntary use, organises it in four functions: Govern, Map, Measure and Manage.SOURCES: NIST · AI Risk Management Framework ↗

AI in participation

Using AI to help people take part in public decisions. An OECD review of June 2026 describes nine types of use, from translation and transcription to moderation and simulation, and warns about skewed data, opaque decisions and simulations that replace real participation.SOURCES: OECD · AI and the Future of Citizen Participation (June 2026) ↗

Not a legal or technical opinion. If a source changes, this page changes with it.

← HOME