AAIA Practice Test 6 — 10 Questions

Question 1

Which THREE components are critical for a robust 'AI Governance' framework?

Accepted Answer

Established AI ethics and compliance policies. Effective governance requires policy, accountability, and continuous monitoring.

Answer

Increasing the size of the marketing team

Answer

Choosing the most expensive cloud platform

Question 2

During an audit of an AI system, you notice that 'Model validation' is performed by the same team that developed the model. Why is this a concern?

Accepted Answer

Lack of independence compromises the integrity of the validation process. Self-validation lacks the independence required to identify errors, bias, or lack of robustness effectively.

Answer

The team might be too friendly with each other

Answer

The validation team might work too slowly

Answer

It is not a concern

Answer

The model will be too accurate

Question 3

When setting up a containerized deployment of an AI model using Docker, what is the primary purpose of a 'multi-stage build'?

Accepted Answer

To reduce the final image size and attack surface. The primary purpose of a multi-stage build in Docker is to reduce the final image size and attack surface by separating the build environment from the runtime environment. Build tools, compilers, and intermediate dependencies are used in an early stage but are not copied into the final image, leaving only the necessary runtime artifacts. This results in a leaner, more secure production image.

Answer

To run multiple versions of the model simultaneously

Answer

To train the model during the container startup

Answer

To automatically scale the deployment based on traffic

Question 4

What is the primary function of an 'AI Model Gateway'?

Accepted Answer

Centralized request management and security. A gateway acts as a security and management layer for AI requests, providing throttling, logging, and authentication.

Answer

Storing model weights

Answer

Training the model

Answer

Labeling data

Question 5

When auditing fairness metrics, what does the 'Demographic Parity' metric measure?

Accepted Answer

Equal positive outcome rates. Demographic Parity (also called statistical parity) requires that the probability of a positive prediction is equal across protected groups — i.e., P(Ŷ=1 | A=0) = P(Ŷ=1 | A=1). It measures outcome-rate parity, not error rates or accuracy. It is one of the simplest group fairness definitions and is often used in hiring, lending, and admissions audits.

Answer

Equal training data representation

Answer

Equal error rates

Answer

Equal accuracy rates

Question 6

To verify the provenance of a model in Google Cloud Vertex AI, which service should the auditor inspect to review the lineage graph of the artifacts?

Accepted Answer

Vertex AI ML Metadata. Vertex AI ML Metadata tracks the lineage of artifacts and executions within the pipeline.

Answer

Vertex AI Feature Store

Answer

Cloud Logging

Answer

Vertex AI Model Registry

Question 7

When auditing fairness using the 'Aequitas' toolkit, what is the first step an auditor should take?

Accepted Answer

Select bias metrics. The first step when auditing fairness using the Aequitas toolkit is to select the bias metrics that will be used to evaluate the model. Aequitas requires the auditor to choose from a set of fairness metrics (e.g., disparate impact, false positive rate parity) based on the context and protected attributes. Only after selecting metrics can the toolkit compute and visualize them. This step ensures the audit is tailored to the specific fairness concerns.

Answer

Encrypt the dataset

Answer

Train a new model

Answer

Delete sensitive attributes

Question 8

What is the goal of an AI 'Rollout' plan?

Accepted Answer

Controlled deployment to users. An AI rollout plan governs the controlled deployment of a model to users, typically through phased stages such as pilot, limited release, and general availability. The goal is to manage risk, gather feedback, and ensure readiness before full-scale exposure. It is fundamentally a deployment governance artifact, not a resource optimization or training activity.

Answer

Reducing disk space

Answer

Deleting logs

Answer

Increasing training time

Question 9

Which TWO are standard practices for securing AI model endpoints?

Accepted Answer

Mutual TLS (mTLS). Mutual TLS (mTLS) is correct because it enforces bidirectional certificate-based authentication, ensuring both the client and the AI model endpoint verify each other's identity before any inference request is processed, which prevents unauthorized or spoofed clients from reaching the model. API Rate Limiting is correct because it caps the number of requests per client or API key over a defined window, mitigating abuse, credential-stuffing, denial-of-service, and cost-exhaustion attacks against expensive model inference calls. Unprotected endpoints do not belong because exposing a model without authentication or encryption allows anyone to invoke it and exfiltrate data or consume resources. Storing logs on public S3 does not belong because publicly readable buckets leak prompts, responses, and potentially sensitive training or user data. Public internet access does not belong because unrestricted exposure of a model endpoint dramatically expands the attack surface instead of restricting access to trusted networks or identities.

Answer

Unprotected endpoints

Answer

Storing logs on public S3

Answer

Public internet access

Question 10

A data scientist proposes using a 'black-box' model for a high-stakes loan approval process. Which action does your AI audit framework require?

Accepted Answer

Mandating the use of XAI (Explainable AI) tools to interpret model decisions. In high-stakes domains like loan approval, regulations (e.g., ECOA, GDPR) and ethical AI principles require that decisions be explainable to affected individuals and auditors. XAI tools (e.g., LIME, SHAP, counterfactual explanations) provide post-hoc interpretability for black-box models, enabling justification of denials and detection of bias. Thus, an AI audit framework would mandate XAI to ensure transparency and accountability, making option A correct.

Answer

Requiring the data scientist to publish the code on an internal repository

Answer

Allowing the model if it achieves 99% accuracy on the test set

Answer

Asking the marketing team to verify the model's fairness