An AI risk team is monitoring a model that predicts equipment failure. They discover that the sensor data feeding the model has inconsistent units across different facilities. What is the most appropriate action?

Switch to a different model that does not rely on sensor data

Standardize the data units across all facilities before using it for monitoring

Use the data as-is and note the inconsistency in the report

Exclude data from facilities with inconsistent units

Which of the following is an example of aligning AI risk reporting with the ERM framework?

Creating a separate risk reporting system that is not shared with other departments

Reporting AI risks only to the AI team and not to senior management

Using the same risk categories and terminology as the rest of the organization

Using unique risk metrics that are not comparable to other enterprise risks

Which of the following is typically included in an AI risk report to the board of directors?

Raw data logs from AI monitoring systems

Detailed technical specifications of every AI model

High-level summary of key AI risks and their alignment with risk appetite

A list of all AI-related incidents with full forensic analysis

Why is data quality important for AI risk metrics?

Because inaccurate data can lead to incorrect risk assessments and decisions

Because data quality is required by all AI regulations

Because data quality reduces the need for human oversight

Because data quality ensures that AI models are always accurate

An AI risk team is setting up monitoring for a customer churn prediction model. They notice that the data feeding the model contains missing values for a significant portion of customers. What is the most important first step to ensure the reliability of the risk metrics?

Switch to a different model that does not require the missing data

Implement data quality checks to identify and address missing values before they affect the metrics

Proceed with monitoring and ignore the missing data

Report the metrics as-is and note the missing data in a footnote

ISACAAdvanced in AI Risk

Domain 3Objective 4

AI Risk Metrics, Monitoring, and Reporting AAIR Practice Questions (Page 4)

Part of the AI Risk Program Management domain, which accounts for 42% of the AAIR exam.

37questions here

8free pages

10concepts

42%of the exam

10 Concepts in This Objective

Define AI risk metrics
Align metrics with risk appetite
Establish monitoring processes
Implement monitoring tools and automation
Develop reporting frameworks
Tailor reports to audience
Ensure data quality and integrity
Review and update metrics
Integrate with enterprise risk management
Document and communicate findings

Questions 16–20

16application · medium
An AI risk team is monitoring a model that predicts equipment failure. They discover that the sensor data feeding the model has inconsistent units across different facilities. What is the most appropriate action?
Select an answer first
17foundation · easy
Which of the following is an example of aligning AI risk reporting with the ERM framework?
Select an answer first
18foundation · easy
Which of the following is typically included in an AI risk report to the board of directors?
Select an answer first
19foundation · easy
Why is data quality important for AI risk metrics?
Select an answer first
20application · medium
An AI risk team is setting up monitoring for a customer churn prediction model. They notice that the data feeding the model contains missing values for a significant portion of customers. What is the most important first step to ensure the reliability of the risk metrics?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to AAIR

Free Basic Practice is a study aid with revealable answers — not a scored exam. Examers.io is independent and not affiliated with or endorsed by ISACA. “AAIR” is a trademark of its owner, used for identification only.