AWS Certified AI Practitioner AIF-C01 Practice Test 42
Question 410
An external chat assistant sometimes returns insulting or hateful language. A pilot uses this approach instead: Use only a vector database index as the content safety filter. Which limitation MOST DIRECTLY explains why the requirement is not met?
- Guardrails can help filter unwanted model inputs and outputs
- Prompt wording alone is not a complete moderation control
- Billing authorization does not inspect generated text
- Retrieval search is not an output toxicity control
Explanation
Correct Answer
D:This is the actual failure mode of the proposed approach: Use only a vector database index as the content safety filter. Retrieval search is not an output toxicity control
Incorrect Answers
A:Guardrails can help filter unwanted model inputs and outputs This explains the recommended design, but it is not the requested diagnosis of Use only a vector database index as the content safety filter.
B:This limitation concerns Rely only on a polite system persona without safety testing; it does not diagnose the proposed Use only a vector database index as the content safety filter approach.
C:This limitation concerns Change IAM billing permissions to remove toxic language; it does not diagnose the proposed Use only a vector database index as the content safety filter approach.
Question 411
A customer-facing medical information assistant may invent unsafe dosage recommendations. After selecting the recommended approach, which evidence MOST DIRECTLY demonstrates that the actual required outcome has been achieved?
- A reviewed design record approves this competing approach: Increase completion length to improve medical authority
- A component test invokes this alternate approach successfully: Disable all source links and let the model reason unaudited
- A reviewed test set rejects invented claims and routes uncertain cases for human review
- A system inventory lists this competing approach as active: Treat any confident generated dosage as approved clinical guidance
Explanation
Correct Answer
C:A reviewed test set rejects invented claims and routes uncertain cases for human review. This checks the business result rather than just a configured feature.
Incorrect Answers
A:Longer advice is not clinical validation Design approval is not evidence that the alternative meets the actual requirement.
B:Removing provenance increases the chance of undetected errors A component invocation does not validate the necessary end-to-end outcome.
D:Model confidence cannot establish medical safety An active setting cannot independently prove the requested business outcome.
Question 412
A bank asks how to communicate performance limitations for a fraud model. After selecting the recommended approach, which evidence MOST DIRECTLY demonstrates that the actual required outcome has been achieved?
- A component test invokes this alternate approach successfully: Publish only the model's hidden weights
- A system inventory lists this competing approach as active: Publish only a marketing statement about AI accuracy
- A reviewed design record approves this competing approach: Provide only a service uptime chart as model quality disclosure
- Users can find intended use, unacceptable uses and known performance limits
Explanation
Correct Answer
D:Users can find intended use, unacceptable uses and known performance limits. This checks the business result rather than just a configured feature.
Incorrect Answers
A:Weights cannot serve as a plain-language limitation record A component invocation does not validate the necessary end-to-end outcome.
B:Marketing copy omits measured limitations and assumptions An active setting cannot independently prove the requested business outcome.
C:Availability does not describe failure modes or bias Design approval is not evidence that the alternative meets the actual requirement.
Question 413
A model offers high accuracy but opaque reasons for high-stakes decisions. During a design review, someone proposes Choose the simplest model regardless of accuracy or safety. Which technical objection is specifically relevant to this proposal?
- Adoption counts do not measure explainability or model reliability
- Opaque gains may not justify lost explainability in regulated decisions
- Pure accuracy can fail explanation obligations
- Interpretability alone does not guarantee acceptable outcomes
Explanation
Correct Answer
D:This is the actual failure mode of the proposed approach: Choose the simplest model regardless of accuracy or safety. Interpretability alone does not guarantee acceptable outcomes
Incorrect Answers
A:This limitation concerns Use model popularity as the risk-adjusted decision; it does not diagnose the proposed Choose the simplest model regardless of accuracy or safety approach.
B:Opaque gains may not justify lost explainability in regulated decisions This explains the recommended design, but it is not the requested diagnosis of Choose the simplest model regardless of accuracy or safety.
C:This limitation concerns Choose highest accuracy without assessing auditability; it does not diagnose the proposed Choose the simplest model regardless of accuracy or safety approach.
Question 414
A firm must disclose that customers are interacting with AI and provide escalation. Which option BEST meets the business requirement?
- Make AI interaction transparent and provide a human handoff path
- Use only automated satisfaction guesses without user recourse
- Require users to accept all outputs without correction
- Hide the model identity to make replies seem human
Explanation
Correct Answer
A:Human-centered design supports informed use and correction of mistakes The specified scenario makes this the most appropriate approach.
Incorrect Answers
B:Predicted sentiment does not let users contest outcomes This prevents the approach from meeting the stated requirement.
C:No contest or escalation path weakens trust This prevents the approach from meeting the stated requirement.
D:Deception removes user choice and feedback This prevents the approach from meeting the stated requirement.
Question 415
An analyst needs a model explanation that business users can understand without revealing training secrets. A pilot uses this approach instead: Reveal private training rows as the explanation. Which limitation MOST DIRECTLY explains why the requirement is not met?
- Useful explainability describes factors affecting predictions in an appropriate form
- Architecture size does not explain a particular prediction
- Exposing source records can violate privacy and still be confusing
- Charges do not establish why the model responded
Explanation
Correct Answer
C:This is the actual failure mode of the proposed approach: Reveal private training rows as the explanation. Exposing source records can violate privacy and still be confusing
Incorrect Answers
A:Useful explainability describes factors affecting predictions in an appropriate form This explains the recommended design, but it is not the requested diagnosis of Reveal private training rows as the explanation.
B:This limitation concerns Use only neural network layer counts for every decision; it does not diagnose the proposed Reveal private training rows as the explanation approach.
D:This limitation concerns Provide a billing receipt instead of decision rationale; it does not diagnose the proposed Reveal private training rows as the explanation approach.
Question 416
A bank asks how to communicate performance limitations for a fraud model. Before approving a rollout, an architect suggests Provide only a service uptime chart as model quality disclosure. Which mismatch should be documented for this precise design?
- Availability does not describe failure modes or bias
- Marketing copy omits measured limitations and assumptions
- Weights cannot serve as a plain-language limitation record
- Transparent documentation lets users understand where output is reliable
Explanation
Correct Answer
A:This is the actual failure mode of the proposed approach: Provide only a service uptime chart as model quality disclosure. Availability does not describe failure modes or bias
Incorrect Answers
B:This limitation concerns Publish only a marketing statement about AI accuracy; it does not diagnose the proposed Provide only a service uptime chart as model quality disclosure approach.
C:This limitation concerns Publish only the model's hidden weights; it does not diagnose the proposed Provide only a service uptime chart as model quality disclosure approach.
D:Transparent documentation lets users understand where output is reliable This explains the recommended design, but it is not the requested diagnosis of Provide only a service uptime chart as model quality disclosure.
Question 417
A firm must disclose that customers are interacting with AI and provide escalation. Before approving a rollout, an architect suggests Require users to accept all outputs without correction. Which mismatch should be documented for this precise design?
- Deception removes user choice and feedback
- No contest or escalation path weakens trust
- Predicted sentiment does not let users contest outcomes
- Human-centered design supports informed use and correction of mistakes
Explanation
Correct Answer
B:This is the actual failure mode of the proposed approach: Require users to accept all outputs without correction. No contest or escalation path weakens trust
Incorrect Answers
A:This limitation concerns Hide the model identity to make replies seem human; it does not diagnose the proposed Require users to accept all outputs without correction approach.
C:This limitation concerns Use only automated satisfaction guesses without user recourse; it does not diagnose the proposed Require users to accept all outputs without correction approach.
D:Human-centered design supports informed use and correction of mistakes This explains the recommended design, but it is not the requested diagnosis of Require users to accept all outputs without correction.
Question 418
An analyst needs a model explanation that business users can understand without revealing training secrets. Before approving a rollout, an architect suggests Provide a billing receipt instead of decision rationale. Which mismatch should be documented for this precise design?
- Charges do not establish why the model responded
- Useful explainability describes factors affecting predictions in an appropriate form
- Architecture size does not explain a particular prediction
- Exposing source records can violate privacy and still be confusing
Explanation
Correct Answer
A:This is the actual failure mode of the proposed approach: Provide a billing receipt instead of decision rationale. Charges do not establish why the model responded
Incorrect Answers
B:Useful explainability describes factors affecting predictions in an appropriate form This explains the recommended design, but it is not the requested diagnosis of Provide a billing receipt instead of decision rationale.
C:This limitation concerns Use only neural network layer counts for every decision; it does not diagnose the proposed Provide a billing receipt instead of decision rationale approach.
D:This limitation concerns Reveal private training rows as the explanation; it does not diagnose the proposed Provide a billing receipt instead of decision rationale approach.
Question 419
A deployed predictor starts receiving different input distributions from the data used for validation. A pilot uses this approach instead: Assume a successful deployment permanently guarantees accuracy. Which limitation MOST DIRECTLY explains why the requirement is not met?
- Deployment state says nothing about future distribution shift
- Uncontrolled retraining can introduce new risk
- Production input changes can degrade predictive behavior
- Storage bytes do not diagnose feature distribution change
Explanation
Correct Answer
A:This is the actual failure mode of the proposed approach: Assume a successful deployment permanently guarantees accuracy. Deployment state says nothing about future distribution shift
Incorrect Answers
B:This limitation concerns Retrain continuously without tracking data or performance changes; it does not diagnose the proposed Assume a successful deployment permanently guarantees accuracy approach.
C:Production input changes can degrade predictive behavior This explains the recommended design, but it is not the requested diagnosis of Assume a successful deployment permanently guarantees accuracy.
D:This limitation concerns Measure only S3 storage growth as model drift; it does not diagnose the proposed Assume a successful deployment permanently guarantees accuracy approach.
Question 420
A GenAI assistant must avoid repeating social security numbers found in user-provided text. Before approving a rollout, an architect suggests Use public CloudTrail events as a repository for raw sensitive prompts. Which mismatch should be documented for this precise design?
- Sampling does not reliably prevent information disclosure
- Unnecessary prompt logging can increase exposure
- Sensitive data handling needs explicit controls rather than model goodwill
- A polite request is not enforceable protection
Explanation
Correct Answer
B:This is the actual failure mode of the proposed approach: Use public CloudTrail events as a repository for raw sensitive prompts. Unnecessary prompt logging can increase exposure
Incorrect Answers
A:This limitation concerns Increase model temperature so identifiers become random; it does not diagnose the proposed Use public CloudTrail events as a repository for raw sensitive prompts approach.
C:Sensitive data handling needs explicit controls rather than model goodwill This explains the recommended design, but it is not the requested diagnosis of Use public CloudTrail events as a repository for raw sensitive prompts.
D:This limitation concerns Rely only on a natural-language request to keep secrets; it does not diagnose the proposed Use public CloudTrail events as a repository for raw sensitive prompts approach.
