Overview
EN ISO/IEC 22989:2023 (ISO/IEC 22989:2022) defines core artificial intelligence concepts and terminology for information technology. The standard establishes a common vocabulary and conceptual framework for AI - covering terms related to data, machine learning, neural networks, trustworthiness, natural language processing, and computer vision. Intended for use across sectors, it supports consistent communication among developers, regulators, standards writers, procurement teams and other stakeholders. EN ISO/IEC 22989:2023 is the European adoption of the international ISO/IEC document and is applicable to all organization types.
Key topics
- Standardized terminology: comprehensive terms and definitions grouped by topic (AI general, data, machine learning, neural networks, trustworthiness, NLP, CV).
- AI concepts: foundational concepts such as agents, knowledge, cognition, symbolic vs subsymbolic approaches, soft computing and genetic algorithms.
- Machine learning taxonomy: supervised, unsupervised, semi‑supervised, reinforcement, transfer learning, training/validation/test data, trained models and retraining.
- Neural networks and algorithms: descriptions of key algorithm families and examples used in AI systems.
- Trustworthiness and governance: concepts like robustness, reliability, resilience, controllability, explainability, transparency, bias and fairness, and verification/validation.
- AI system life cycle: a life‑cycle model and staged processes for AI system development, deployment and maintenance.
- Application domains: terminology and concepts for NLP and computer vision to align cross‑discipline communication.
Practical applications
- Standards development: provides the foundational vocabulary to draft consistent, interoperable AI standards and technical specifications.
- Procurement and contracts: clarifies expectations by using agreed definitions for capabilities, requirements and metrics.
- Regulation and policy: assists regulators and legal teams in interpreting laws and guidelines with consistent meanings for AI terms.
- Engineering and testing: helps engineers, QA and validation teams align on lifecycle activities, data roles, model definitions and trustworthiness criteria.
- Education and training: useful for curricula and corporate training to ensure consistent understanding of AI concepts.
Who should use this standard
- Standards bodies and technical committees
- AI system designers, architects and developers
- Regulators, auditors and compliance teams
- Procurement officers and contract managers
- Academic researchers, trainers and policy makers
Related standards
EN ISO/IEC 22989:2023 is part of the broader ISO/IEC JTC 1 work on AI. Use it alongside sector‑specific AI standards and other ISO/IEC AI guidance to ensure consistent terminology across documents.
Keywords: ISO/IEC 22989, EN ISO/IEC 22989:2023, AI concepts and terminology, artificial intelligence terminology, AI lifecycle, machine learning, trustworthiness, natural language processing, computer vision.