AIF-C01 Guidelines for Responsible AI Practice Test 2 — 15 Questions

Question 1

A team is deploying a regression model for loan approval. To ensure transparency for regulators, they need to explain individual predictions. Which interpretability method can provide local explanations by approximating the model with a simpler surrogate?

Accepted Answer

LIME. Option C is correct because LIME (Local Interpretable Model-agnostic Explanations) generates local explanations by fitting a simpler model around each prediction. SHAP (option A) is also local but uses Shapley values; partial dependence (option B) is global; permutation importance (option D) is global feature importance.

Answer

SHAP values

Answer

Partial dependence plots

Answer

Permutation feature importance

Question 2

A healthcare company must train a model on sensitive patient data while complying with privacy regulations. They want to add noise to the training process to prevent re-identification. Which technique should they implement?

Accepted Answer

Differential privacy. Option B is correct because differential privacy injects controlled noise into the training algorithm to protect individual records. k-anonymity (A) focuses on generalization, not noise injection. Homomorphic encryption (C) allows computation on ciphertext but is not typically used during training. Federated learning (D) trains on decentralized data but does not inherently provide differential privacy guarantees.

Answer

k-anonymity

Answer

Federated learning

Answer

Homomorphic encryption

Question 3

After deploying a model, a company notices that the distribution of the input features has shifted compared to the training data. Which feature of Amazon SageMaker Model Monitor can alert them to this change?

Accepted Answer

Data quality monitoring. Option C is correct because SageMaker Model Monitor's Data Quality monitoring tracks distributions of input features and can alert on drift. Bias drift (A) is specific to demographic groups. Model quality (B) tracks prediction accuracy. Feature importance drift (D) is not a standard monitoring type.

Answer

Model quality monitoring

Answer

Bias drift monitoring

Answer

Feature importance drift

Question 4

A company uses Amazon SageMaker Ground Truth to label a dataset for a binary classifier. To reduce labeling bias, which workforce configuration is most appropriate?

Accepted Answer

Private workforce of domain experts. Option D is correct because using an AWS-managed workforce (Mechanical Turk) often involves multiple workers per item, but for sensitive tasks, a private workforce of domain experts ensures consistency and reduces bias. Public (A) may introduce uncontrolled bias. Vendor (B) is not a built-in type. Automatic (C) is not a workforce type.

Answer

Automatic labeling with Active Learning

Answer

Public workforce with no qualification

Answer

Vendor managed workforce

Question 5

A machine learning team is building a credit risk model and discovers that the training data has a significant imbalance in loan approval rates between two demographic groups. They decide to reweight the training samples using a preprocessing technique. Which SageMaker Clarify feature can help compute the appropriate sample weights to achieve demographic parity?

Accepted Answer

Clarify preprocessing (reweighting). Option A is correct because SageMaker Clarify's preprocessing transforms include a reweighting method that assigns weights to instances to adjust for fairness. Post-training (B, C) are not preprocessing. Monitoring (D) is post-deployment.

Answer

Clarify post-training bias metrics

Answer

Model Monitor bias drift

Answer

Clarify explainability (SHAP)

Question 6

Which of the following is a key principle of responsible AI according to AWS?

Accepted Answer

Transparency. Option B is correct because AWS emphasizes transparency as a key principle of responsible AI, alongside fairness, accountability, privacy, and robustness. Profitability (A), speed (C), and complexity (D) are not principles.

Answer

Complexity

Answer

Speed

Answer

Profitability

Question 7

A company wants to ensure accountability for its machine learning models by tracking all changes to the model and its training data. Which AWS feature should they use?

Accepted Answer

AWS CloudTrail. Option B is correct because AWS CloudTrail records API calls for auditing and accountability. SageMaker experiments (A) track training parameters but not all data changes. CodeCommit (C) is for source code, not model changes. CloudWatch (D) monitors metrics but does not track changes comprehensively.

Answer

Amazon CloudWatch

Answer

Amazon SageMaker Experiments

Answer

AWS CodeCommit

Question 8

A security team is concerned about adversarial attacks on their image classification model deployed on Amazon SageMaker. They want to test robustness against carefully crafted inputs that cause misclassification. What approach should they use?

Accepted Answer

Generating adversarial examples using SageMaker Clarify. Option D is correct because adversarial validation involves generating adversarial examples using attacks like FGSM to test model robustness. Model Monitor (A) does not test adversarial robustness. Data augmentation (B) improves generalization but is not a test. A/B testing (C) compares models, not robustness.

Answer

Data augmentation on the training set

Answer

A/B testing between two similar models

Answer

SageMaker Model Monitor with adversarial drift

Question 9

Which TWO actions should a data scientist take to evaluate fairness of a binary classification model using Amazon SageMaker Clarify? (Choose two.)

Accepted Answer

Use post-training bias metrics like Difference in Positive Proportions. Options A and D are correct. SageMaker Clarify computes pre-training bias metrics (like Class Imbalance) and post-training metrics (like Difference in Positive Proportions) to assess fairness. Option B is for explainability, not bias. Option C is for unlabeled data, not model evaluation. Option E is a general practice but not specific to Clarify fairness evaluation.

Answer

Ensure the training dataset is balanced by resampling

Answer

Generate SHAP values for feature importance

Answer

Run a data quality monitoring job on unlabeled data

Question 10

Which THREE considerations are important when implementing responsible AI for a production NLP system? (Choose three.)

Accepted Answer

Continuously monitor model outputs for bias and drift. Options A, B, and D are correct. In production, continuous monitoring is vital (A), bias evaluation should be part of the MLOps pipeline (B), and model cards document model details (D). Option C is about code security, not responsible AI. Option E is specific to medical devices, not generally required.

Answer

Obtain FDA approval for the model

Answer

Apply encryption at rest for all training code

Question 11

Which TWO techniques provide interpretability for machine learning models at a local (per-prediction) level? (Choose two.)

Accepted Answer

SHAP values. Options A and C are correct. LIME and SHAP are local interpretability methods. Option B (Permutation importance) is global. Option D (Partial dependence) is global. Option E (Confusion matrix) is a performance metric, not interpretability.

Answer

Partial dependence plots

Answer

Confusion matrix

Answer

Permutation feature importance

Question 12

Refer to the exhibit. A data scientist runs an Amazon SageMaker Clarify bias analysis on a binary classifier. The pre-training ClassImbalance is 1.5 and the post-training DPPL is 0.15. What should the data scientist conclude?

Accepted Answer

The data has a mild class imbalance, but the model shows a noticeable bias in predictions.. Option B is correct. A ClassImbalance of 1.5 indicates the majority class is 1.5x the minority, mild imbalance. A DPPL of 0.15 indicates a 15% difference in positive prediction rates between groups, which is a significant fairness concern. Option A misinterprets both; C is wrong because bias is present; D confuses the metrics.

Answer

The data is highly imbalanced and the model is unbiased.

Answer

The pre-training metric indicates a fairness issue, but the post-training metric is acceptable.

Answer

The data is perfectly balanced and the model is fair.

Question 13

Refer to the exhibit. A team is creating an IAM policy for a SageMaker notebook user. The user needs to access training data in an S3 bucket and create models. Which responsible AI concern is most relevant to this policy?

Accepted Answer

The policy grants overly broad permissions, violating the principle of least privilege.. Option C is correct. The policy grants broad access (sagemaker:CreateModel and sagemaker:CreateNotebookInstance on all resources) without restrictions. This could allow a user to create models using any data or expose the notebook. The least privilege principle is violated, leading to potential unintended model creation or data exposure. Options A and B are less directly related; D is about explainability.

Answer

The policy does not enforce encryption for the notebook.

Answer

The policy does not restrict which S3 buckets the user can read.

Answer

The policy does not include a condition for model explainability.

Question 14

Refer to the exhibit. A developer is reviewing CloudWatch Logs for a deployed model and notices the same input appears multiple times with slightly different probabilities. What responsible AI concern does this pattern suggest?

Accepted Answer

The model is not robust; it produces inconsistent predictions for the same input.. Option B is correct. Repeated identical inputs with different predictions indicate model instability (lack of robustness). This could be due to randomness in the model or adversarial conditions. Option A is irrelevant; C is possible but not the primary concern; D is about data drift, but input is same.

Answer

The model is overfitting to the training data.

Answer

The model is exhibiting bias against a demographic group.

Answer

The input data is drifting from the training distribution.

Question 15

A company uses Amazon Rekognition for facial analysis. They want to ensure the model doesn't exhibit bias based on skin tone. What should they do?

Accepted Answer

Ensure the training dataset includes diverse skin tones. Option D is correct: Training on diverse data reduces bias. Option A is wrong: Network size does not address bias. Option B is wrong: Data augmentation does not guarantee diversity. Option C is wrong: Pre-trained models may have inherent bias.

Answer

Apply data augmentation to increase dataset size

Answer

Use a larger neural network

Answer

Use a pre-trained model from AWS Marketplace