ISO IEC 5338-2023 PDF

Full title and description

ISO/IEC 5338:2023 — Information technology — Artificial intelligence — AI system life cycle processes. This International Standard defines a structured set of life‑cycle processes and associated concepts for describing, managing and improving AI systems (including machine‑learning and heuristic systems) throughout their development, acquisition, operation and maintenance.

Abstract

This document specifies processes that support the definition, control, management, execution and continual improvement of AI systems across their life cycle. It aligns with and extends established software and system life‑cycle frameworks (ISO/IEC/IEEE 15288 and ISO/IEC/IEEE 12207) by adding AI‑specific processes and concepts drawn from related AI standards. It is intended for use when developing, acquiring or integrating AI system elements and when traditional software or systems elements are present the corresponding ISO/IEEE software/system life‑cycle processes can be used to implement them.

General information

• Status: Published
• Publication date: December 2023
• Publisher: ISO / IEC (International Standard published under ISO/IEC JTC 1/SC 42)
• ICS / categories: 35.020 (Information technology)
• Edition / version: Edition 1 (2023-12)
• Number of pages: 39

Scope

The standard defines a set of life‑cycle processes and related concepts specifically for AI systems based on machine learning and heuristic approaches. It provides process descriptions that can be applied at the project or organizational level to define, control, manage, execute and improve AI systems through their life‑cycle stages, and describes how AI life‑cycle processes interact with traditional software and system life‑cycle processes (ISO/IEC/IEEE 12207 and 15288).

Key topics and requirements

• Definition and structuring of AI system life‑cycle processes (planning, development/acquisition, integration, deployment, operation, maintenance, retirement).
• Alignment and mapping with existing software/system life‑cycle standards (ISO/IEC/IEEE 12207 and 15288).
• Inclusion of AI‑specific processes and concepts (model/data management, training/validation cycles, monitoring, model updates and governance).
• Requirements for process tailoring, roles and responsibilities, and process outcomes to support consistent management of AI systems.
• Support for verification, validation, risk management, and continual improvement across AI life cycle stages.

Typical use and users

Used by organizations and projects that develop, acquire or operate AI systems: systems and software engineers, AI/ML engineers, project and program managers, procurement teams, quality and assurance functions, auditors, and regulators seeking a process framework for lifecycle governance and integration with established software/system engineering practices.

Related standards

This standard explicitly references and aligns with ISO/IEC/IEEE 15288 and ISO/IEC/IEEE 12207 and draws on AI‑specific work such as ISO/IEC 22989 and ISO/IEC 23053. It is part of the suite of AI standards developed under ISO/IEC JTC 1/SC 42.

Keywords

AI life cycle, life‑cycle processes, AI governance, model management, data management, verification, validation, monitoring, ISO/IEC JTC 1/SC 42, machine learning lifecycle.

FAQ

Q: What is this standard?

A: ISO/IEC 5338:2023 is an international standard that defines life‑cycle processes for AI systems to guide their development, acquisition, operation and improvement.

Q: What does it cover?

A: It covers a structured set of processes and associated concepts for AI systems (particularly machine‑learning and heuristic systems), including how those processes map to existing software and system life‑cycle standards.

Q: Who typically uses it?

A: Systems and software engineers, AI/ML practitioners, project managers, procurement and quality teams, auditors and regulators — essentially any stakeholder responsible for delivering or overseeing AI systems and their life‑cycle governance.

Q: Is it current or superseded?

A: It is a current, published International Standard (published December 2023, Edition 1). Users should check national/adopted versions or corrigenda for any updates since publication.

Q: Is it part of a series?

A: It is part of the ISO/IEC JTC 1/SC 42 body of AI standards and is intended to be used alongside related standards such as ISO/IEC 22989, ISO/IEC 23053 and the ISO/IEC/IEEE life‑cycle standards (15288, 12207).

Q: What are the key keywords?

A: AI life cycle, life‑cycle processes, model management, data governance, verification and validation, monitoring, ISO/IEC JTC 1/SC 42.