WHO dossier/Glossary
Part 02 · Glossary
Definitions become useful when they shape review.
The paper defines AI-related terms so policy teams can discuss tools, techniques, outputs, and governance without collapsing every system into the same risk category.
Key terms
Plain-language decoding.
source vocabularyMachine capability for intelligent tasks
For iFeed use, this means naming the task: prediction, classification, generation, search, optimization, ranking, or decision support.
Systems that create new outputs
Policy users should separate generated text from source evidence and record how generated material was checked.
Evidence-informed policy-making
Policy decisions use research, data, practice insight, context, and values; AI can support this but cannot replace judgement.
Reflection
Vocabulary as a control.
team useBefore using AI in an evidence workflow, teams should name the technique, the output, the decision influence, the evidence source, and the human review point. Shared language is the first layer of governance.