ISO/IEC 22989:2022, the vocabulary
22989 sets the vocabulary for AI and sets out the ideas used across the field. Its terms apply wherever the FDIS uses them, so a local definition must not contradict it.
The requirement in plain English
22989 sets the vocabulary for AI and sets out the ideas used across the field. Its terms apply wherever the FDIS uses them, so a local definition must not contradict it.
What the public record says
The public view gives the scope, the terms and the introduction. The scope says it supports the drafting of other standards and exchange between interested parties.
Why it matters to you
The FDIS does not define AI, AI system or machine learning itself; it borrows them. That makes 22989 the place where a definitional argument is settled. It also means a change in 22989 flows into the FDIS reading. The practical work is small but real: take the twenty terms your robustness evidence uses and check each against 22989, then record where you rely on it. A term that is not in 22989 and not defined in the FDIS has no normative home, and a reviewer will notice.
Evidence a reviewer asks for
- The glossary mapping each term to 22989, with the clause and the date read.
- A list of local definitions that differ from 22989, with the reason.
- The reading date and edition in the method statement.
Checks to run this month
- Which terms in the evidence pack are defined in 22989?
- Is any local definition in conflict?
- Who owns the terminology decision?
Dates
22989 published in 2022. FDIS registered 2026-08-21.
Cross-references
- Clause 3, Terms and definitions
- ISO/IEC 23053:2022, the system framework
- Part 1, the overview of robustness assessment
What this page does not state
The public view carries the scope and terms, not the full concept clauses.