AI Guide For Boards in Singapore

AI GUIDE FOR BOARDS IN SINGAPORE

Section V: Organisational Resilience and Capability

11.4 Model Lifecycle Governance

approval processes. Clear ownership and accountability are important, as is continuous performance monitoring. Management should also dene escalation pathways and retirement or rollback procedures. Without appropriate lifecycle governance, organisations may face model dri and degraded performance, hidden bias, poor explainability, operational instability, and reduced auditability and regulatory defensibility.

Eective AI governance depends not only on the technology itself, but also on how data and AI models are managed across their lifecycle. AI models are dynamic systems that evolve over time as data, operational conditions and user behaviour change. Boards should therefore ensure the organisation maintains Model Lifecycle Governance across the full lifecycle of AI systems. Eective lifecycle governance helps ensure AI systems remain reliable, secure, explainable, auditable and aligned with business objectives and risk appetite.

Boards should ensure that systems are in place to explain decisions made by critical AI systems.

Boards should expect management to establish formal model validation and

Model Lifecycle Governance

CONTINUOUS LIFECYCLE MANAGEMENT PROCESS: FEEDBACK & IMPROVEMENT LOOP

1

2

3

4

5

6

DESIGN DEVELOP VALIDATE

DEPLOY MONITOR RETIRE

FEEDBACK & IMPROVEMENT

99

Made with FlippingBook Digital Proposal Creator