Materiality Assessment
Risk-based determination of an AI application's significance.
What it triggers
- In Singapore, the Monetary Authority of Singapore's Artificial Intelligence Model Risk Management – Observations from a Thematic Review sets out a materiality assessment to determine the level of controls and governance proportionate to the AI application's potential impact.source
- Under the Artificial Intelligence Model Risk Management – Observations from a Thematic Review, a materiality assessment evaluates an AI use-case across multiple risk dimensions to identify the appropriate controls and governance.source
Definition
A structured evaluation of an AI use-case across multiple risk dimensions (financial, operational, compliance, model integrity, data protection, reputational and conduct risk) to determine the level of controls and governance proportionate to the application's potential impact.
Not to be confused with
A material AI system refers to an AI application already determined to have significant impact or regulatory relevance, while materiality assessment is the process to evaluate and determine that significance.
Material impact describes specific customer or market-facing effects causing measurable harm, whereas materiality assessment is the broader evaluation process across multiple risk dimensions to gauge overall significance.
Questions and answers
What is involved in a materiality assessment for AI under Singapore guidelines?
A materiality assessment is a structured evaluation of an AI use-case across multiple risk dimensions including financial, operational, compliance, model integrity, data protection, reputational and conduct risk. It determines the level of controls and governance proportionate to the application's potential impact.
Does the MAS information paper require materiality assessments for AI?
The Monetary Authority of Singapore's Artificial Intelligence Model Risk Management – Observations from a Thematic Review is a non-binding guideline that sets supervisory expectations for managing AI risks, including through materiality assessments.
How does materiality assessment differ from identifying a material AI system?
Materiality assessment is the risk-based process to evaluate an AI application's significance across various dimensions, while a material AI system is the outcome: an AI application deemed to have significant impact or regulatory relevance.
Related Terms
materiality
Impact on customers, operations, financial soundness or market integrity....
Material AI Portfolio
Collection of AI systems material to an entity's operations or risk profile....
material impact
Customer or market-facing effects causing measurable harm....
Risk classification (of AI applications)
Assessment of AI applications by materiality and risk factors....
Material AI system
An AI application with significant impact or regulatory relevance....