ISO/IEC 22989:2022

ISO/IEC 22989 establishes standardized terminology and conceptual frameworks for artificial intelligence, facilitating consistent communication across industries. This internationally recognized standard provides organizations with clear definitions for AI components, processes, and systems, supporting governance frameworks and promoting responsible AI development through shared understanding of key concepts. Additionally, it incorporates crucial ethical considerations that guide responsible practices.

The Evolution of AI Terminology Standardization

The field of artificial intelligence has experienced exponential growth in recent years, integrating diverse disciplines from computer science and data analytics to cognitive psychology and ethics. This convergence has created a complex landscape where terminology inconsistencies can lead to misunderstandings, compliance challenges, and implementation barriers.

ISO/IEC 22989, first published in 2022, addressed the need for a unified language in AI discourse. Developed through rigorous collaboration among international experts from over 25 countries, this standard applies across cultural, technical, and regulatory contexts, reflecting the array of technical aspects and ethical considerations crucial for AI development.

In early 2025, the standard underwent its first major review, incorporating feedback from practical implementations and aligning with evolving global AI regulations. This review process reinforced the standard's role as a foundational element in the broader AI governance ecosystem, critical as organizations face increasing compliance under frameworks like the EU AI Act.

The Structure and Scope of ISO/IEC 22989

The standard organizes key sections to build a comprehensive terminology framework:

Foundational Concepts

ISO/IEC 22989 defines essential AI concepts serving as building blocks for more complex terminology. These include:

• Artificial intelligence: Capability of an engineered system to acquire, process, and apply knowledge and skills
• AI system: System utilizing AI technologies
• Machine learning: Process using computational algorithms that improve automatically through experience

According to the National Institute of Standards and Technology, consistent application of these definitions has reduced cross-departmental confusion in AI implementations by approximately 37%.

Technical Framework Terminology

The standard details terminology for technical components and processes, relevant for algorithm development and data management:

• Training data: Dataset for training AI systems
• Inference: Process of applying an AI system to produce results
• Neural network: Computational model inspired by biological neural networks
• Reinforcement learning: Machine learning where an agent learns by maximizing rewards through actions

AI System Properties and Characteristics

ISO/IEC 22989 defines terms related to AI system qualities:

• Explainability: Extent to which an AI system's mechanics can be explained in human terms
• Robustness: Ability to maintain performance under varying conditions
• Transparency: Property ensuring information about development, deployment, operation, and outcomes is available to relevant stakeholders

These definitions have become increasingly important as transparency in AI has emerged as a competitive advantage for organizations aiming to build trust.

2025 Updates: Strengthening the Terminology Framework

In March 2025, ISO released a supplementary technical report expanding terminology related to AI assurance mechanisms, coinciding with Nemko's commitment to strengthening capabilities for AI assurance across industries.

The update introduces terms reflecting technological advances:

• Foundation model terminology: Definitions for large, adaptable AI models
• Human-AI collaboration: Updated terminology for interfaces between humans and AI systems
• Algorithmic impact assessment: Standardized terminology for evaluating AI algorithms' effects

According to the International Standards Organization, these additions reflect AI technologies' rapid evolution and broadening applications across sectors.

The Critical Importance of Standardized AI Terminology

Enhanced Organizational Alignment

Standardized terminology bridges departmental silos, enhancing communication among technical teams, compliance officers, and leadership. This alignment is crucial for:

• Strategic planning: Defining AI initiatives and roadmaps
• Risk assessments: Evaluating potential issues across AI applications
• Cross-functional collaborations: Facilitating AI projects across domains

A 2025 survey by the IEEE Standards Association found organizations implementing standardized AI terminology reported 42% faster resolution of disagreements.

Streamlined Compliance Processes

Standardized terminology is essential for efficient compliance as AI regulations proliferate globally. ISO/IEC 22989 provides vocabulary for:

• Documentation requirements: Consistent AI system descriptions for regulatory submissions
• Certification processes: Clear communication with certification bodies
• Audit preparation: Consistent records for audits

Organizations seeking AI regulatory compliance benefit significantly from this standardized vocabulary.

Improved Stakeholder Communication

Clear terminology enhances communication with external stakeholders:

• Customers: Understanding AI capabilities and limitations
• Partners: Integrating with AI systems
• Investors: Assessing AI maturity and risk
• Regulators: Evaluating compliance

Implementation Approaches for ISO/IEC 22989

Organizations can adopt ISO/IEC 22989 through practical approaches:

Terminology Integration Strategy

A phased approach includes:

1. Assessment: Evaluate current terminology usage
2. Gap analysis: Identify inconsistencies with ISO/IEC 22989
3. Prioritization: Target high-impact terms
4. Implementation plan: Develop a timeline
5. Training: Educate staff on definitions

Documentation and Knowledge Management

Effective implementation requires robust documentation:

• Terminology database: Repository of standardized terms
• Style guides: Reflect standardized terminology in writing
• Templates: Incorporate consistent AI terminology
• Review processes: Ensure terminology consistency

Cross-Standard Harmonization

ISO/IEC 22989 functions within a broader ecosystem of AI standards, including:

• ISO/IEC 23053 for AI trustworthiness assessment
• ISO/IEC TR 24028 for AI trustworthiness
• ISO/IEC 42001 for AI management systems

Organizations implementing these standards benefit from the consistent terminology provided by ISO/IEC 22989.

Practical Applications of ISO/IEC 22989

AI Procurement Processes

Standardized terminology enables:

• Clear specification of requirements
• Precise evaluation criteria
• Consistent contractual language

AI Development Lifecycle

Throughout development, standardized terminology supports:

• Requirements gathering
• Design documentation
• Testing protocols
• Deployment guidelines

Risk Management Frameworks

ISO/IEC 22989 enhances risk management by:

• Providing definitions for risk categories
• Identifying AI-specific risks
• Describing mitigation measures

The Strategic Value of Terminology Standardization

As AI transforms industries, ISO/IEC 22989 provides an essential foundation for effective communication, governance, and compliance. Its comprehensive framework enables organizations to:

• Communicate about AI internally and externally
• Align with regulatory requirements
• Build trust through transparent AI practices
• Engage in AI governance effectively

By adopting ISO/IEC 22989, organizations demonstrate commitment to ethical AI development, gaining a strategic advantage in an increasingly regulated landscape. The investment in terminology standardization yields returns through improved efficiency, enhanced compliance readiness, and strengthened stakeholder relationships.