Fundamentals of AI and ML

Question 1 · 1.1

A bank wants a model that flags card transactions as fraudulent or legitimate. It has five years of transactions, each marked by investigators as fraud or not fraud.

Which type of machine learning fits this data?

• ASupervised learning
• BUnsupervised learning
• CReinforcement learning
• DGenerative AI

Question 2 · 1.1

A media company runs a model each night that tags every video uploaded that day with topic labels. Nobody needs the tags until the next morning.

Which inference approach is most cost-effective?

• ABatch inference
• BReal-time inference
• CServerless inference
• DStreaming inference

Question 3 · 1.1

Which statement correctly describes the relationship between these terms?

• ADeep learning is a subset of machine learning, which is a subset of AI.
• BMachine learning is a subset of deep learning.
• CGenerative AI and machine learning are unrelated fields.
• DEvery AI system uses deep learning.

Question 4 · 1.1

A model scores 99% accuracy on the data it was trained on but 62% on new customer data.

What does this most likely indicate?

• AThe model is overfit.
• BThe model is underfit.
• CThe model has a good fit.
• DThe model needs real-time inference.

Question 5 · 1.1

An insurer is considering three systems: one that predicts claim cost, one that drafts letters to claimants, and one that receives a claim, checks the policy, requests missing documents, and schedules an assessor without a human stepping in.

Which TWO statements are correct? (Select TWO.) (Select 2.)

• AThe letter-drafting system is generative AI.
• BThe end-to-end claims system is agentic AI.
• CThe claim-cost predictor is generative AI because it produces a number.
• DAll three systems are agentic because they run automatically.
• EThe letter-drafting system must be trained with reinforcement learning.

Question 6 · 1.2

A travel company wants its mobile app to read booking confirmations aloud to visually impaired customers in a natural-sounding voice.

Which AWS service should it use?

• AAmazon Polly
• BAmazon Transcribe
• CAmazon Comprehend
• DAmazon Lex

Question 7 · 1.2

An online shop has a year of purchase history but no predefined customer segments. Marketing wants to discover natural groups of customers with similar buying behaviour.

Which technique fits?

• AClustering
• BClassification
• CRegression
• DReinforcement learning

Question 8 · 1.2

A payroll team asks for an AI model to calculate each employee's statutory tax deduction from salary and published tax bands.

What is the best recommendation?

• AImplement the published rules directly in code; ML is not appropriate.
• BTrain a regression model on past payslips.
• CUse a foundation model with a prompt describing the tax bands.
• DUse Amazon Comprehend to read the tax rules.

Question 9 · 1.2

A regulated lender must give every declined applicant the main reasons for the decision. It has ten years of labelled, tabular application data.

Which approach is MOST appropriate?

• AA traditional, interpretable supervised model trained on the tabular data
• BA large language model prompted to approve or decline each application
• CClustering applicants into groups
• DAmazon Polly

Question 10 · 1.2

A support centre records customer calls in Spanish. It wants English transcripts and then a sentiment score for each call.

Which THREE services, used together, meet the requirement? (Select THREE.) (Select 3.)

• AAmazon Transcribe
• BAmazon Translate
• CAmazon Comprehend
• DAmazon Polly
• EAmazon Lex

Question 11 · 1.3

A data scientist combines raw purchase timestamps into new variables such as “purchases in the last 7 days” and “average basket value”, then drops columns that add no signal.

Which stage of the ML lifecycle is this?

• AFeature engineering
• BModel monitoring
• CBusiness goal identification
• DHyperparameter tuning

Question 12 · 1.3

A fraud model for a payments company must catch as many fraudulent transactions as possible. Every flagged transaction is reviewed by an analyst before any action is taken. Fraud is about 0.5% of transactions.

Which metric should the team prioritise?

• ARecall
• BAccuracy
• CPrecision
• DCost per user

Question 13 · 1.3

Six months after launch, a demand-forecasting model's error has steadily increased. Customer buying patterns have changed since the training data was collected.

Which MLOps practice addresses this?

• AMonitoring the model and re-training it on recent data
• BSwitching from real-time to batch inference
• CIncreasing the endpoint instance size
• DRewriting the business goal

Question 14 · 1.3

A start-up wants to add a text-summarisation feature next month. It has no ML engineers and does not want to manage servers or GPUs.

Which approach fits best?

• AUse a foundation model through Amazon Bedrock's managed API
• BPre-train a new language model in SageMaker AI
• CSelf-host an open-source model on EC2 GPU instances
• DUse Amazon Polly

Question 15 · 1.3

Which TWO are business metrics rather than model performance metrics? (Select TWO.) (Select 2.)

• AReturn on investment (ROI)
• BCost per user
• CF1 score
• DRecall
• EPrecision