AWS Certified AI Practitioner AIF-C01 Practice Test 27
Question 260
A retailer needs a recurring branded tone across thousands of responses based on a stable style guide. Before approving a rollout, an architect suggests Rely only on an embedding vector index to enforce every phrasing choice. Which mismatch should be documented for this precise design?
- Weight adaptation can improve consistent behavior for a stable specialized style
- Retrieval supplies context but does rather than persistently alter weights
- Quota changes do not teach the desired tone
- Similarity search alone is not a trained response style
Explanation
Correct Answer
D:This is the actual failure mode of the proposed approach: Rely only on an embedding vector index to enforce every phrasing choice. Similarity search alone is not a trained response style
Incorrect Answers
A:Weight adaptation can improve consistent behavior for a stable specialized style This explains the recommended design, but it is not the requested diagnosis of Rely only on an embedding vector index to enforce every phrasing choice.
B:This limitation concerns Assume RAG always changes the model's learned style permanently; it does not diagnose the proposed Rely only on an embedding vector index to enforce every phrasing choice approach.
C:This limitation concerns Use a token quota increase as the style training method; it does not diagnose the proposed Rely only on an embedding vector index to enforce every phrasing choice approach.
Question 261
A policy knowledge graph contains entities and connections that must be explored with contextual search. Which option BEST meets the business requirement?
- Use plain token counting to answer multi-hop relationship queries
- Use a sentiment score as the graph traversal algorithm
- Assume vector-only lookup automatically reasons over every graph edge
- Consider graph-aware retrieval such as Amazon Neptune Analytics where it fits
Explanation
Correct Answer
D:Graph relationships can complement retrieval when connections matter The specified scenario makes this the most appropriate approach.
Incorrect Answers
A:Token counts do not encode entity connections This prevents the approach from meeting the stated requirement.
B:Sentiment analysis does not follow entity relationships This prevents the approach from meeting the stated requirement.
C:Vector similarity alone does not ensure traversal of required relationships This prevents the approach from meeting the stated requirement.
Question 262
An HR assistant's answers need recently revised procedures without changing the base model's style. Before approving a rollout, an architect suggests Store updated files but never retrieve them during answering. Which mismatch should be documented for this precise design?
- Retrieval updates facts without requiring repeated model training
- Unused documents cannot ground generation
- Sampling changes cannot load new documents
- Full pretraining for frequent policy updates is excessive
Explanation
Correct Answer
B:This is the actual failure mode of the proposed approach: Store updated files but never retrieve them during answering. Unused documents cannot ground generation
Incorrect Answers
A:Retrieval updates facts without requiring repeated model training This explains the recommended design, but it is not the requested diagnosis of Store updated files but never retrieve them during answering.
C:This limitation concerns Change temperature and assume it refreshes factual weights; it does not diagnose the proposed Store updated files but never retrieve them during answering approach.
D:This limitation concerns Retrain the entire foundation model on every HR policy edit; it does not diagnose the proposed Store updated files but never retrieve them during answering approach.
Question 263
A contract model must infer the desired classification format from several labeled examples supplied in a prompt. During a design review, someone proposes Use a random list of unrelated paragraphs as examples. Which technical objection is specifically relevant to this proposal?
- Unrelated demonstrations may confuse the mapping
- Few-shot examples illustrate how to map new inputs to target outputs
- No demonstrations makes it zero-shot
- Billing settings do not provide task examples
Explanation
Correct Answer
A:This is the actual failure mode of the proposed approach: Use a random list of unrelated paragraphs as examples. Unrelated demonstrations may confuse the mapping
Incorrect Answers
B:Few-shot examples illustrate how to map new inputs to target outputs This explains the recommended design, but it is not the requested diagnosis of Use a random list of unrelated paragraphs as examples.
C:This limitation concerns Call a prompt with no examples few-shot; it does not diagnose the proposed Use a random list of unrelated paragraphs as examples approach.
D:This limitation concerns Replace examples with a larger token billing limit; it does not diagnose the proposed Use a random list of unrelated paragraphs as examples approach.
Question 264
A summarization model must capture key ideas from reference summaries. During a design review, someone proposes Use AWS KMS encryption status as a summary metric. Which technical objection is specifically relevant to this proposal?
- Training loss does not directly measure summary quality against a reference
- Fast output may still omit or invent facts
- ROUGE compares overlap with reference text but misses some semantic quality
- Key configuration does not evaluate language output
Explanation
Correct Answer
D:This is the actual failure mode of the proposed approach: Use AWS KMS encryption status as a summary metric. Key configuration does not evaluate language output
Incorrect Answers
A:This limitation concerns Use only training loss to score final user-facing summaries; it does not diagnose the proposed Use AWS KMS encryption status as a summary metric approach.
B:This limitation concerns Treat low latency alone as summary faithfulness; it does not diagnose the proposed Use AWS KMS encryption status as a summary metric approach.
C:ROUGE compares overlap with reference text but misses some semantic quality This explains the recommended design, but it is not the requested diagnosis of Use AWS KMS encryption status as a summary metric.
Question 265
A company needs to compare two Bedrock FMs on accuracy, safety and human preference. During a design review, someone proposes Select by context length alone. Which technical objection is specifically relevant to this proposal?
- Cheap unsafe output can fail the business requirement
- Bedrock evaluation supports task-specific model comparisons
- Brand identity does not prove task performance
- Capacity does not establish safety or correctness
Explanation
Correct Answer
D:This is the actual failure mode of the proposed approach: Select by context length alone. Capacity does not establish safety or correctness
Incorrect Answers
A:This limitation concerns Use cost only and omit quality measures; it does not diagnose the proposed Select by context length alone approach.
B:Bedrock evaluation supports task-specific model comparisons This explains the recommended design, but it is not the requested diagnosis of Select by context length alone.
C:This limitation concerns Choose by provider brand alone; it does not diagnose the proposed Select by context length alone approach.
Question 266
A policy knowledge graph contains entities and connections that must be explored with contextual search. Before approving a rollout, an architect suggests Assume vector-only lookup automatically reasons over every graph edge. Which mismatch should be documented for this precise design?
- Graph relationships can complement retrieval when connections matter
- Sentiment analysis does not follow entity relationships
- Vector similarity alone does not ensure traversal of required relationships
- Token counts do not encode entity connections
Explanation
Correct Answer
C:This is the actual failure mode of the proposed approach: Assume vector-only lookup automatically reasons over every graph edge. Vector similarity alone does not ensure traversal of required relationships
Incorrect Answers
A:Graph relationships can complement retrieval when connections matter This explains the recommended design, but it is not the requested diagnosis of Assume vector-only lookup automatically reasons over every graph edge.
B:This limitation concerns Use a sentiment score as the graph traversal algorithm; it does not diagnose the proposed Assume vector-only lookup automatically reasons over every graph edge approach.
D:This limitation concerns Use plain token counting to answer multi-hop relationship queries; it does not diagnose the proposed Assume vector-only lookup automatically reasons over every graph edge approach.
Question 267
A team repeatedly edits instructions used by three Bedrock applications and needs a common revision history. Which option BEST meets the business requirement?
- Write prompts into CloudTrail as the only prompt-template system
- Store only untracked text snippets in personal chat histories
- Change model temperature to create an audit history
- Use Amazon Bedrock Prompt Management to version and manage reusable prompts
Explanation
Correct Answer
D:A managed prompt resource supports controlled prompt iteration and versions The specified scenario makes this the most appropriate approach.
Incorrect Answers
A:Audit logs record API activities rather than manage reusable prompts This prevents the approach from meeting the stated requirement.
B:Ad hoc prompts lack shared version governance This prevents the approach from meeting the stated requirement.
C:Sampling settings do not track who approved a prompt This prevents the approach from meeting the stated requirement.
Question 268
An image model should avoid including brand logos in generated artwork. Which option BEST meets the business requirement?
- Assume only a model's training history prevents every unwanted brand
- Specify the unwanted content explicitly in negative prompting or exclusions
- Mention the prohibited logo as a desired featured element
- Set the output token count to one as a logo filter
Explanation
Correct Answer
B:Negative constraints can steer generation away from named elements The specified scenario makes this the most appropriate approach.
Incorrect Answers
A:Training provenance alone does not enforce prompt-specific exclusions This prevents the approach from meeting the stated requirement.
C:Positive instructions can encourage the excluded object This prevents the approach from meeting the stated requirement.
D:Length settings are not semantic content filters This prevents the approach from meeting the stated requirement.
Question 269
A policy knowledge graph contains entities and connections that must be explored with contextual search. A pilot uses this approach instead: Use plain token counting to answer multi-hop relationship queries. Which limitation MOST DIRECTLY explains why the requirement is not met?
- Vector similarity alone does not ensure traversal of required relationships
- Sentiment analysis does not follow entity relationships
- Token counts do not encode entity connections
- Graph relationships can complement retrieval when connections matter
Explanation
Correct Answer
C:This is the actual failure mode of the proposed approach: Use plain token counting to answer multi-hop relationship queries. Token counts do not encode entity connections
Incorrect Answers
A:This limitation concerns Assume vector-only lookup automatically reasons over every graph edge; it does not diagnose the proposed Use plain token counting to answer multi-hop relationship queries approach.
B:This limitation concerns Use a sentiment score as the graph traversal algorithm; it does not diagnose the proposed Use plain token counting to answer multi-hop relationship queries approach.
D:Graph relationships can complement retrieval when connections matter This explains the recommended design, but it is not the requested diagnosis of Use plain token counting to answer multi-hop relationship queries.
Question 270
A feedback program collects human preference pairs for generated customer responses. A pilot uses this approach instead: Use unlabeled random characters as preference rankings. Which limitation MOST DIRECTLY explains why the requirement is not met?
- Uninterpretable data cannot provide reliable preference signals
- Encryption protects data but does not teach response preferences
- Preference comparisons can help train models toward human-valued behavior
- Token volume does not express which answer humans prefer
Explanation
Correct Answer
A:This is the actual failure mode of the proposed approach: Use unlabeled random characters as preference rankings. Uninterpretable data cannot provide reliable preference signals
Incorrect Answers
B:This limitation concerns Replace preference review with only storage encryption; it does not diagnose the proposed Use unlabeled random characters as preference rankings approach.
C:Preference comparisons can help train models toward human-valued behavior This explains the recommended design, but it is not the requested diagnosis of Use unlabeled random characters as preference rankings.
D:This limitation concerns Assume a token-count report is an alignment label; it does not diagnose the proposed Use unlabeled random characters as preference rankings approach.
