AWS Certified AI Practitioner AIF-C01 Practice Test 30
Question 290
A company needs to compare two Bedrock FMs on accuracy, safety and human preference. After selecting the recommended approach, which evidence MOST DIRECTLY demonstrates that the actual required outcome has been achieved?
- A system inventory lists this competing approach as active: Choose by provider brand alone
- Comparative evaluation uses representative prompts and recorded scores
- A component test invokes this alternate approach successfully: Select by context length alone
- A reviewed design record approves this competing approach: Use cost only and omit quality measures
Explanation
Correct Answer
B:Comparative evaluation uses representative prompts and recorded scores. This checks the business result rather than just a configured feature.
Incorrect Answers
A:Brand identity does not prove task performance An active setting cannot independently prove the requested business outcome.
C:Capacity does not establish safety or correctness A component invocation does not validate the necessary end-to-end outcome.
D:Cheap unsafe output can fail the business requirement Design approval is not evidence that the alternative meets the actual requirement.
Question 291
A policy knowledge graph contains entities and connections that must be explored with contextual search. 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: Assume vector-only lookup automatically reasons over every graph edge
- A test query recovers the intended entities and linked relationships
- A component test invokes this alternate approach successfully: Use a sentiment score as the graph traversal algorithm
- A system inventory lists this competing approach as active: Use plain token counting to answer multi-hop relationship queries
Explanation
Correct Answer
B:A test query recovers the intended entities and linked relationships. This checks the business result rather than just a configured feature.
Incorrect Answers
A:Vector similarity alone does not ensure traversal of required relationships Design approval is not evidence that the alternative meets the actual requirement.
C:Sentiment analysis does not follow entity relationships A component invocation does not validate the necessary end-to-end outcome.
D:Token counts do not encode entity connections An active setting cannot independently prove the requested business outcome.
Question 292
A travel assistant must answer in several regional languages with citations to local laws. A pilot uses this approach instead: Choose any text model solely by English benchmarks. Which limitation MOST DIRECTLY explains why the requirement is not met?
- Large completions are not proof of language coverage
- Cost savings cannot compensate for unusable translations
- English performance does not prove regional language accuracy
- Language quality and factual grounding are explicit selection criteria
Explanation
Correct Answer
C:This is the actual failure mode of the proposed approach: Choose any text model solely by English benchmarks. English performance does not prove regional language accuracy
Incorrect Answers
A:This limitation concerns Pick by longest generated output without multilingual testing; it does not diagnose the proposed Choose any text model solely by English benchmarks approach.
B:This limitation concerns Choose cheapest without examining linguistic accuracy; it does not diagnose the proposed Choose any text model solely by English benchmarks approach.
D:Language quality and factual grounding are explicit selection criteria This explains the recommended design, but it is not the requested diagnosis of Choose any text model solely by English benchmarks.
Question 293
A retailer needs a recurring branded tone across thousands of responses based on a stable style guide. Which option BEST meets the business requirement?
- Assume RAG always changes the model's learned style permanently
- Consider supported fine-tuning or customization using curated examples
- Rely only on an embedding vector index to enforce every phrasing choice
- Use a token quota increase as the style training method
Explanation
Correct Answer
B:Weight adaptation can improve consistent behavior for a stable specialized style The specified scenario makes this the most appropriate approach.
Incorrect Answers
A:Retrieval supplies context but does not persistently alter weights This prevents the approach from meeting the stated requirement.
C:Similarity search alone is not a trained response style This prevents the approach from meeting the stated requirement.
D:Quota changes do not teach the desired tone This prevents the approach from meeting the stated requirement.
Question 294
A multi-step support agent appears fluent but regularly fails to complete the requested customer workflow. Before approving a rollout, an architect suggests Treat configured tools as proof every action executed. Which mismatch should be documented for this precise design?
- Tool availability is not successful authorized execution
- Retrieval quality alone is not end-to-end task success
- Longer responses can mask unfinished work
- Agent value depends on actual workflow outcomes, not only attractive prose
Explanation
Correct Answer
A:This is the actual failure mode of the proposed approach: Treat configured tools as proof every action executed. Tool availability is not successful authorized execution
Incorrect Answers
B:This limitation concerns Measure retrieval relevance without tracking whether actions finished; it does not diagnose the proposed Treat configured tools as proof every action executed approach.
C:This limitation concerns Score only conversational word count; it does not diagnose the proposed Treat configured tools as proof every action executed approach.
D:Agent value depends on actual workflow outcomes, not only attractive prose This explains the recommended design, but it is not the requested diagnosis of Treat configured tools as proof every action executed.
Question 295
A customer service bot needs accurate answers from an internal policy collection that changes weekly. 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: Increase temperature until the model recalls current internal clauses
- Generated answers cite relevant freshly indexed policy passages
- A system inventory lists this competing approach as active: Fine-tune on one static snapshot as the only update mechanism
- A reviewed design record approves this competing approach: Rely only on public pretraining for private documents
Explanation
Correct Answer
B:Generated answers cite relevant freshly indexed policy passages. This checks the business result rather than just a configured feature.
Incorrect Answers
A:Randomness does not provide current policy evidence A component invocation does not validate the necessary end-to-end outcome.
C:Static weights may not reflect weekly document changes An active setting cannot independently prove the requested business outcome.
D:Pretrained data does not contain confidential current policies by default Design approval is not evidence that the alternative meets the actual requirement.
Question 296
A summarization model must capture key ideas from reference summaries. After selecting the recommended approach, which evidence MOST DIRECTLY demonstrates that the actual required outcome has been achieved?
- ROUGE and human factual checks jointly improve against reviewed reference summaries
- A reviewed design record approves this competing approach: Treat low latency alone as summary faithfulness
- A component test invokes this alternate approach successfully: Use AWS KMS encryption status as a summary metric
- A system inventory lists this competing approach as active: Use only training loss to score final user-facing summaries
Explanation
Correct Answer
A:ROUGE and human factual checks jointly improve against reviewed reference summaries. This checks the business result rather than just a configured feature.
Incorrect Answers
B:Fast output may still omit or invent facts Design approval is not evidence that the alternative meets the actual requirement.
C:Key configuration does not evaluate language output A component invocation does not validate the necessary end-to-end outcome.
D:Training loss does not directly measure summary quality against a reference An active setting cannot independently prove the requested business outcome.
Question 297
A travel assistant must answer in several regional languages with citations to local laws. Which option BEST meets the business requirement?
- Pick by longest generated output without multilingual testing
- Choose any text model solely by English benchmarks
- Evaluate multilingual capability and source grounding alongside price
- Choose cheapest without examining linguistic accuracy
Explanation
Correct Answer
C:Language quality and factual grounding are explicit selection criteria The specified scenario makes this the most appropriate approach.
Incorrect Answers
A:Large completions are not proof of language coverage This prevents the approach from meeting the stated requirement.
B:English performance does not prove regional language accuracy This prevents the approach from meeting the stated requirement.
D:Cost savings cannot compensate for unusable translations This prevents the approach from meeting the stated requirement.
Question 298
A specialized science assistant requires substantial domain-language familiarity absent from generic pretraining. 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: Treat a new user interface as domain adaptation
- A reviewed design record approves this competing approach: Assume retrieval alone always yields the desired base domain language behavior
- A system inventory lists this competing approach as active: Use prompt caching as the only way to teach unfamiliar specialized language
- Evaluation shows domain terminology handled correctly on independent test sets
Explanation
Correct Answer
D:Evaluation shows domain terminology handled correctly on independent test sets. This checks the business result rather than just a configured feature.
Incorrect Answers
A:Presentation changes do not alter model language knowledge A component invocation does not validate the necessary end-to-end outcome.
B:RAG helps facts but may not substitute for deep domain adaptation Design approval is not evidence that the alternative meets the actual requirement.
C:Caching reused prefixes does not train model weights An active setting cannot independently prove the requested business outcome.
Question 299
A feedback program collects human preference pairs for generated customer responses. During a design review, someone proposes Assume a token-count report is an alignment label. Which technical objection is specifically relevant to this proposal?
- Preference comparisons can help train models toward human-valued behavior
- Encryption protects data but does not teach response preferences
- Uninterpretable data cannot provide reliable preference signals
- Token volume does not express which answer humans prefer
Explanation
Correct Answer
D:This is the actual failure mode of the proposed approach: Assume a token-count report is an alignment label. Token volume does not express which answer humans prefer
Incorrect Answers
A:Preference comparisons can help train models toward human-valued behavior This explains the recommended design, but it is not the requested diagnosis of Assume a token-count report is an alignment label.
B:This limitation concerns Replace preference review with only storage encryption; it does not diagnose the proposed Assume a token-count report is an alignment label approach.
C:This limitation concerns Use unlabeled random characters as preference rankings; it does not diagnose the proposed Assume a token-count report is an alignment label approach.
Question 300
A healthcare team is curating fine-tuning examples and many contain sensitive patient identifiers. A pilot uses this approach instead: Use every raw patient record without consent assessment. Which limitation MOST DIRECTLY explains why the requirement is not met?
- Encryption does not replace lawful data-use authorization
- Unvetted sensitive training data creates privacy and compliance risks
- Training data governance and privacy matter before model adaptation
- Weights do not by themselves guarantee training privacy
Explanation
Correct Answer
B:This is the actual failure mode of the proposed approach: Use every raw patient record without consent assessment. Unvetted sensitive training data creates privacy and compliance risks
Incorrect Answers
A:This limitation concerns Assume file encryption alone establishes the right to use all medical records; it does not diagnose the proposed Use every raw patient record without consent assessment approach.
C:Training data governance and privacy matter before model adaptation This explains the recommended design, but it is not the requested diagnosis of Use every raw patient record without consent assessment.
D:This limitation concerns Treat the model's hidden parameters as proof no patient data can leak; it does not diagnose the proposed Use every raw patient record without consent assessment approach.
