Decoder. plain-English AI glossary

Transparency

● Core

Also called AI Transparency

Being open about how a model was built, what it can do, and where it falls short — so people can judge it.

Think of it like

A nutrition label on food: you may still eat it, but you get to see what’s inside first.

Example

A provider publishes a model card documenting training data sources, evaluation results, known limitations, and intended uses.

How it actually works

Transparency spans documentation (model cards, data statements), disclosure of limitations and risks, and sometimes access for external auditors. It is a precondition for accountability — you cannot hold a system responsible for behavior no one can inspect. The tension is that full openness can also aid misuse, so providers balance disclosure against safety.

For product teams

What lets users, regulators, and auditors trust your system with eyes open rather than blind.

For engineers

Document data, capabilities, evals, and limits (e.g. model cards); it is the substrate auditing and accountability build on.

Related

Read anything AI without the jargon

Look up any term in plain English, or save terms as you read with the free Chrome extension.

Open DecoderAdd to Chrome