Free GARP RAI Exam Questions

Become GARP Certified with updated RAI exam questions and correct answers

Page: 1 / 66

Total 330 Questions | Updated On: Mar 13, 2026

Add To Cart

Question 1

An insurance company uses two different AI models for policy approval, one for Group A and another for Group B. The confusion matrices for each group are as follows:Group A Confusion Matrix:True Positives: 50False Positives: 10False Negatives: 40True Negatives: 90Group B Confusion Matrix:True Positives: 30False Positives: 20False Negatives: 70True Negatives: 80What fairness technique should the company use to balance the performance between the two groups?

A : Equalize the number of true positives for both groups.

B : Adjust thresholds to achieve similar false positive rates for both groups.

C : Apply a higher penalty for false negatives in Group A.

D : Implement demographic parity by ensuring equal acceptance rates for both groups.

Answer: D

Question 2

A financial institution is implementing an AI model that uses proxies for loan approval decisions. Which of the following is a potential risk associated with using proxies?

A : Proxies might result in simplified metrics that don’t align with the actual objective.

B : Proxies require extensive computational resources.

C : Proxies ensure a direct alignment with the model’s intended objective.

D : Proxies make the model fully explainable to external auditors.

Answer: A

Question 3

A bank is developing its model risk governance framework to address ML/AI models. What foundational approach should the bank take?

A : Create an entirely new framework specific to AI/ML.

B : Build upon its existing model risk framework while adapting to new challenges from AI/ML.

C : Exclude traditional models from the risk framework.

D : Focus solely on AI/ML ethical aspects without considering statistical issues.

Answer: B

Question 4

A financial institution is creating a model inventory. Which of the following should be included to ensure comprehensive governance?

A : Only the model’s purpose and validation report.

B : Model owner, but not developer information, for clarity.

C : A description of model assumptions but not its risk classification.

D : Model owner, purpose, methodology, assumptions, limitations, and risk tier.

Answer: D

Question 5

An insurance company is using a neural network for classifying claims as "Fraudulent" or "Non-Fraudulent." They decide to use ReLU (Rectified Linear Unit) as the activation function for the hidden layers. What is the primary purpose of using an activation function like ReLU in this neural network?

A : To scale the output between 0 and 1

B : To introduce non-linearity

C : To calculate weights in the hidden layers

D : To set bias terms to zero

Answer: B

Page: 1 / 66

Total 330 Questions | Updated On: Mar 13, 2026

Add To Cart