AAISM Practice Test 8 — 10 Questions

Question 1

When a model exhibits 'Hallucinations', what is the best strategy to improve its groundedness and security?

Accepted Answer

Implement Retrieval-Augmented Generation (RAG). RAG (Retrieval-Augmented Generation) forces the model to base its answers on specific, provided documents, reducing hallucination risks.

Answer

Restrict access to the model to root users only

Answer

Disable logging to hide hallucinations

Answer

Increase the model's training data volume

Question 2

When assessing AI third-party vendor risk, which document is most useful for understanding the vendor's data handling practices?

Accepted Answer

The Data Processing Agreement (DPA).. The Data Processing Agreement (DPA) outlines exactly how data is used, stored, and protected.

Answer

The user interface mockups.

Answer

The company's marketing brochure.

Answer

The vendor's quarterly earnings statement.

Question 3

What is 'Model Stealing' as a risk?

Accepted Answer

When an attacker extracts the model's parameters/weights.. Model stealing is the attempt to gain unauthorized access to the underlying weights of a model, often through side-channel attacks or memory scraping.

Answer

When the model is used for a purpose other than intended.

Answer

When the training data is corrupted.

Answer

When a developer quits and takes the code.

Question 4

A model is vulnerable to 'Evasion Attacks' using FGSM (Fast Gradient Sign Method). Which architectural change most effectively increases robustness?

Accepted Answer

Applying Defensive Distillation.. FGSM uses the model's gradients to craft perturbations; defensive distillation or adversarial training helps mask these gradients.

Answer

Increasing the learning rate during training.

Answer

Switching to a larger neural network.

Answer

Implementing input denoising.

Question 5

When evaluating an AI model for production, what does the 'F1 Score' tell you from a risk management perspective?

Accepted Answer

The balance between model precision and recall.. The F1 score provides a balanced view of precision and recall, helping assess the risk of both false positives and false negatives.

Answer

The latency of the model in milliseconds.

Answer

The total number of parameters in the model.

Answer

The total cost of training the model.

Question 6

Which THREE technical configurations help mitigate the risk of 'Model Exfiltration'?

Accepted Answer

Strict IAM policies limiting access to the model weight files.. Model exfiltration can be hindered by restricting access to the model, monitoring egress traffic, and obfuscating model responses.

Answer

Increasing the number of public endpoints.

Answer

Using clear text for all model logs.

Question 7

What is the primary goal of 'Model Monitoring' in the context of risk management?

Accepted Answer

To detect performance degradation, drift, and anomalous behavior.. Continuous monitoring detects anomalies, performance degradation, and drift, which are critical triggers for risk interventions.

Answer

To reduce the cost of GPU usage.

Answer

To block all internet traffic.

Answer

To update the marketing website automatically.

Question 8

In the context of AI governance, what is the primary role of a 'Model Card'?

Accepted Answer

Provide transparency on model capabilities and limitations. A Model Card is a documentation artifact that provides transparency into the model's intended use, limitations, and training data.

Answer

Grant IAM access to the model file

Answer

Enable encryption for the model weights

Answer

Automatically patch the model software

Question 9

Which TWO techniques help mitigate 'Model Poisoning'?

Accepted Answer

Data sanitization and filtering of training inputs.. Defensive techniques like data sanitization and robust aggregation algorithms help mitigate poisoning attempts.

Answer

Increasing the number of neurons in the model.

Answer

Moving the model to a public cloud environment.

Answer

Reducing the number of training epochs.

Question 10

What is 'AI Red Teaming'?

Accepted Answer

Proactive adversarial testing to identify model weaknesses.. AI Red Teaming is a structured process of testing an AI system's safety and security by attempting to find vulnerabilities through adversarial attacks.

Answer

The marketing department's review of AI ads.

Answer

An automated script that changes the model color.

Answer

The process of painting red marks on the server racks.