ISO/IEC 22989:2022

Overview

ISO/IEC 22989:2022 - "Information technology - Artificial intelligence - Artificial intelligence concepts and terminology" - defines a common vocabulary and explains core concepts across the AI domain. The standard establishes terminology for AI and describes concepts used by standards bodies, developers, regulators and stakeholders to improve clarity, consistency and interoperability. It is applicable to all types of organizations (commercial, government, not-for-profit) and is intended to support the development of other standards and facilitate communication among diverse interested parties.

Key topics

This standard systematically covers concepts and terms, including:

• Core AI terminology and high-level AI concepts (agents, knowledge, cognition)
• Data-related terms and roles (training, validation, test data)
• Machine learning concepts and subtypes (supervised, unsupervised, semi-supervised, reinforcement, transfer learning)
• Neural networks and example algorithms (neural nets, Bayesian networks, decision trees, SVM)
• Trustworthiness taxonomy (robustness, reliability, resilience, controllability, explainability, bias/fairness, transparency)
• Domain-specific terminology (natural language processing, computer vision)
• AI system lifecycle (inception, design & development, verification & validation, deployment, operation & monitoring, continuous validation, retirement)
• Stakeholder roles (AI provider/producer/customer/partner/subject, authorities)
• AI ecosystem and functional overview (data, learning, prediction → decision → action), plus IoT, cyber-physical systems, big data, cloud/edge concerns

Practical applications

ISO/IEC 22989 is a foundational reference for:

• Creating consistent technical documentation, specifications and API contracts
• Drafting policy, procurement requirements and contractual clauses that reference standardized AI terms
• Developing AI governance, risk and compliance (GRC) frameworks and audit checklists
• Aligning multidisciplinary teams (engineers, legal, product, ethics reviewers) around shared definitions
• Designing curricula and training materials on AI concepts and lifecycle
• Assisting standards writers and regulators when producing domain-specific AI standards

Who should use this standard

• AI system architects and ML engineers seeking common vocabulary
• Standards developers and technical committees drafting AI-related norms
• Policy makers, regulators and procurement officers writing requirements or guidance
• Legal, compliance and risk teams assessing AI governance and accountability
• Educators and trainers developing AI courses or certification programs

Related standards

ISO/IEC 22989 is part of the broader ISO/IEC AI standards landscape and is intended to be used alongside other AI-specific standards addressing trustworthiness, safety, data practices and evaluation methods to ensure consistent interpretation and application of AI concepts.

Keywords: ISO/IEC 22989, AI terminology, artificial intelligence concepts, AI lifecycle, machine learning, trustworthiness, neural networks, NLP, computer vision.