AWS Certified AI Practitioner AIF-C01 Practice Test 1
Question 1
A streaming company has no audience-category labels but wants to discover patterns in viewing behavior. During a design review, someone proposes Use time-series forecasting to produce only next week's total traffic. Which technical objection is specifically relevant to this proposal?
- Clustering can discover similar users without predefined class labels
- Forecasting aggregates volume rather than discovering user groups
- Language translation does not group viewing patterns by behavior
- A supervised classifier requires an appropriate labeled target
Explanation
Correct Answer
B:This is the actual failure mode of the proposed approach: Use time-series forecasting to produce only next week's total traffic. Forecasting aggregates volume rather than discovering user groups
Incorrect Answers
A:Clustering can discover similar users without predefined class labels This explains the recommended design, but it is not the requested diagnosis of Use time-series forecasting to produce only next week's total traffic.
C:This limitation concerns Use translation of descriptions as the segmentation algorithm; it does not diagnose the proposed Use time-series forecasting to produce only next week's total traffic approach.
D:This limitation concerns Train supervised classification using nonexistent segment labels; it does not diagnose the proposed Use time-series forecasting to produce only next week's total traffic approach.
Question 2
A binary disease-screening model misses many true positive cases although its overall accuracy looks high. Before approving a rollout, an architect suggests Use token output cost as the medical false-negative measure. Which mismatch should be documented for this precise design?
- A continuous regression metric does not directly show missed positive classifications
- Inference spending does not count clinical positives missed
- Recall measures recovered actual positives and reveals the misses that accuracy can obscure
- Accuracy can appear high when negative cases dominate
Explanation
Correct Answer
B:This is the actual failure mode of the proposed approach: Use token output cost as the medical false-negative measure. Inference spending does not count clinical positives missed
Incorrect Answers
A:This limitation concerns Use mean absolute error as the primary binary confusion-matrix statistic; it does not diagnose the proposed Use token output cost as the medical false-negative measure approach.
C:Recall measures recovered actual positives and reveals the misses that accuracy can obscure This explains the recommended design, but it is not the requested diagnosis of Use token output cost as the medical false-negative measure.
D:This limitation concerns Use only accuracy without inspecting class prevalence; it does not diagnose the proposed Use token output cost as the medical false-negative measure approach.
Question 3
A demand model performs nearly perfectly on training records but poorly on a new seasonal holdout set. A pilot uses this approach instead: Report training-set accuracy alone as proof of future accuracy. Which limitation MOST DIRECTLY explains why the requirement is not met?
- Training fit does not measure performance on unseen seasonal samples
- A large training-to-holdout gap indicates memorization or fit that does not generalize
- Deployment completion is independent of predictive generalization
- More optimization can increase overfitting when generalization is already weak
Explanation
Correct Answer
A:This is the actual failure mode of the proposed approach: Report training-set accuracy alone as proof of future accuracy. Training fit does not measure performance on unseen seasonal samples
Incorrect Answers
B:A large training-to-holdout gap indicates memorization or fit that does not generalize This explains the recommended design, but it is not the requested diagnosis of Report training-set accuracy alone as proof of future accuracy.
C:This limitation concerns Skip validation because the model finished deploying successfully; it does not diagnose the proposed Report training-set accuracy alone as proof of future accuracy approach.
D:This limitation concerns Increase the number of training epochs without checking holdout loss; it does not diagnose the proposed Report training-set accuracy alone as proof of future accuracy approach.
Question 4
A contact center wants to classify whether a customer review expresses positive, neutral, or negative sentiment. Before approving a rollout, an architect suggests Use Amazon Textract to infer the writer's attitude from an existing text string. Which mismatch should be documented for this precise design?
- Comprehend can analyze sentiment in natural-language documents
- Textract extracts document content and is not the sentiment classifier
- Translation changes language rather than assigning sentiment categories
- Speech synthesis renders audio but is not the text sentiment analysis step
Explanation
Correct Answer
B:This is the actual failure mode of the proposed approach: Use Amazon Textract to infer the writer's attitude from an existing text string. Textract extracts document content and is not the sentiment classifier
Incorrect Answers
A:Comprehend can analyze sentiment in natural-language documents This explains the recommended design, but it is not the requested diagnosis of Use Amazon Textract to infer the writer's attitude from an existing text string.
C:This limitation concerns Use Amazon Translate as the sentiment classifier; it does not diagnose the proposed Use Amazon Textract to infer the writer's attitude from an existing text string approach.
D:This limitation concerns Use Amazon Polly to derive sentiment from text-to-speech audio; it does not diagnose the proposed Use Amazon Textract to infer the writer's attitude from an existing text string approach.
Question 5
A retailer needs to estimate each next-day order's delivery cost in dollars from historical shipment features. Which option BEST meets the business requirement?
- Use a reinforcement agent rewarded for opening support tickets
- Use a multiclass classifier that predicts only high or low risk
- Use k-means to form unlabeled customer segments
- Use supervised regression to predict a numeric delivery cost
Explanation
Correct Answer
D:A continuous target such as cost calls for regression using known historical values The specified scenario makes this the most appropriate approach.
Incorrect Answers
A:Ticket-opening rewards do not model the labeled delivery-cost relationship This prevents the approach from meeting the stated requirement.
B:A class label loses the requested specific dollar estimate This prevents the approach from meeting the stated requirement.
C:Customer grouping does not predict a numeric shipping target This prevents the approach from meeting the stated requirement.
Question 6
A transit application needs to read an already generated arrival announcement aloud for accessibility. A pilot uses this approach instead: Use Amazon Transcribe to turn the arrival text into an audio clip. Which limitation MOST DIRECTLY explains why the requirement is not met?
- Polly creates spoken audio from supplied text
- Translate produces translated text rather than spoken synthesis
- Transcribe converts speech to text in the opposite direction
- Recognizing entities does not render speech
Explanation
Correct Answer
C:This is the actual failure mode of the proposed approach: Use Amazon Transcribe to turn the arrival text into an audio clip. Transcribe converts speech to text in the opposite direction
Incorrect Answers
A:Polly creates spoken audio from supplied text This explains the recommended design, but it is not the requested diagnosis of Use Amazon Transcribe to turn the arrival text into an audio clip.
B:This limitation concerns Use Amazon Translate as the text-to-speech engine; it does not diagnose the proposed Use Amazon Transcribe to turn the arrival text into an audio clip approach.
D:This limitation concerns Use Amazon Comprehend entity detection as the speech generator; it does not diagnose the proposed Use Amazon Transcribe to turn the arrival text into an audio clip approach.
Question 7
An insurance provider receives recordings of approved claims interviews and needs searchable text transcripts. Which option BEST meets the business requirement?
- Use Amazon Transcribe to convert the speech recordings to text
- Use Amazon Polly for the audio-to-text conversion
- Use Amazon Translate without first obtaining a transcript
- Use Amazon Rekognition labels to extract spoken words from audio
Explanation
Correct Answer
A:Automatic speech recognition converts audio speech into written transcript content The specified scenario makes this the most appropriate approach.
Incorrect Answers
B:Polly generates audio from text rather than transcribing speech This prevents the approach from meeting the stated requirement.
C:Translation alone does not perform the requested speech recognition This prevents the approach from meeting the stated requirement.
D:Rekognition image/video analysis is not a general speech transcription service This prevents the approach from meeting the stated requirement.
Question 8
An accounts-payable department needs to extract invoice fields and table cells from scanned PDF documents. Which option BEST meets the business requirement?
- Use Amazon Polly to read the scanned PDF as an invoice database
- Use Amazon Textract forms and tables extraction
- Use Amazon Rekognition general object labels as the invoice parser
- Use Amazon Translate to extract source PDF table coordinates
Explanation
Correct Answer
B:Textract is designed to extract text and structured fields or tables from document images The specified scenario makes this the most appropriate approach.
Incorrect Answers
A:Speech synthesis does not parse scanned invoice forms This prevents the approach from meeting the stated requirement.
C:Visual object labels do not reconstruct invoice form keys and table cell structure This prevents the approach from meeting the stated requirement.
D:Language translation does not identify document tables and fields This prevents the approach from meeting the stated requirement.
Question 9
A data analyst receives transaction tables, customer chat text and store-camera photos for one research initiative. After selecting the recommended approach, which evidence MOST DIRECTLY demonstrates that the actual required outcome has been achieved?
- The pipeline routes tables, text and images through compatible processing before model use
- A component test invokes this alternate approach successfully: Treat every numeric transaction column as a video frame
- A system inventory lists this competing approach as active: Classify every input as tabular merely because it is stored in S3
- A reviewed design record approves this competing approach: Assume labeled and unlabeled mean the same thing as structured and unstructured
Explanation
Correct Answer
A:The pipeline routes tables, text and images through compatible processing before model use. This checks the business result rather than just a configured feature.
Incorrect Answers
B:Numeric table columns are not image or video pixels A component invocation does not validate the necessary end-to-end outcome.
C:Object-storage location does not make image pixels or natural text tabular An active setting cannot independently prove the requested business outcome.
D:Target-label availability and data structure are separate properties Design approval is not evidence that the alternative meets the actual requirement.
Question 10
An operations team must process a million stored images overnight rather than respond to user uploads instantly. During a design review, someone proposes Train a new model for each uploaded image. Which technical objection is specifically relevant to this proposal?
- Audio conversion does not produce requested visual predictions
- Re-training per input is not normal inference and adds avoidable cost
- Bulk offline prediction need not maintain low per-request interactive latency
- Interactive calls add unnecessary per-request pressure when offline processing is allowed
Explanation
Correct Answer
B:This is the actual failure mode of the proposed approach: Train a new model for each uploaded image. Re-training per input is not normal inference and adds avoidable cost
Incorrect Answers
A:This limitation concerns Replace image inference with an audio conversion service; it does not diagnose the proposed Train a new model for each uploaded image approach.
C:Bulk offline prediction need not maintain low per-request interactive latency This explains the recommended design, but it is not the requested diagnosis of Train a new model for each uploaded image.
D:This limitation concerns Choose only synchronous low-latency requests for each image; it does not diagnose the proposed Train a new model for each uploaded image approach.
