AWS Certified AI Practitioner (AIF-C01) practice questions and exam guide

What is bias in a machine learning model?

1. Random noise added to inputs during training
2. The model's inability to fit non-linear data
3. Systematic error that unfairly favors certain outcomes
4. The variance of predictions across training runs
5. The regularization penalty applied to model weights

Answer: C — Systematic error that unfairly favors certain outcomes

A) Data augmentation, not bias. B) Underfitting from limited capacity. C) Correct — systematic skew (statistical or societal) toward certain outcomes. D) Variance, not bias. E) L1/L2 regularization.

Which BEST describes Amazon SageMaker?

1. A managed service for hosting relational databases
2. A managed service for foundation model chat only
3. A managed service for building serverless web APIs
4. A managed service for end-to-end ML development
5. A managed service for extract-transform-load pipelines

Answer: D — A managed service for end-to-end ML development

A) RDS. B) That's Bedrock. C) API Gateway/Lambda. D) Correct — build, train, tune, deploy, and monitor ML models. E) AWS Glue.

Which BEST describes the difference between training and inference?

1. Training fits weights on data; inference uses the trained model to predict
2. Training predicts on new data; inference updates weights in place
3. Both stages update weights, only the compute type differs
4. Both stages only run on CPUs, never on GPUs or accelerators

Answer: A — Training fits weights on data; inference uses the trained model to predict

A) Correct — canonical distinction. B) Reversed. C) Inference is weight-frozen. D) Accelerators are common at both stages.

Prompt engineering refers to:

1. Fine-tuning weights on a supervised dataset
2. Compressing model artifacts for edge deployment
3. Designing inputs to steer the model's outputs
4. Auto-scaling the endpoint based on token throughput

Answer: C — Designing inputs to steer the model's outputs

A) Fine-tuning. B) Model compression. C) Correct — role, context, examples, format, and constraints. D) Auto-scaling policy.

Which BEST describes a foundation model?

1. A small classifier tuned to one specific task
2. A rules-based knowledge graph used for reasoning
3. A pretrained deep model adaptable to many tasks
4. A statistical technique for regression on tabular data
5. A cloud service that hosts container-based workloads

Answer: C — A pretrained deep model adaptable to many tasks

A) Foundation models are broad, not task-specific. B) Describes symbolic AI. C) Correct — trained on broad data, adapted (fine-tune/prompt/RAG) to many downstream tasks. D) Describes classical stats. E) Describes ECS/EKS/Fargate.

Which BEST describes prompt engineering?

1. Continuing pre-training on general web text
2. Updating the transformer's weight matrices offline
3. Increasing the model's context window at runtime
4. Redesigning the tokenizer to shrink vocabulary size
5. Designing inputs, examples, and constraints to guide outputs

Answer: E — Designing inputs, examples, and constraints to guide outputs

A) Continued pre-training. B) Fine-tuning. C) Not a technique — the window is architectural. D) Tokenizer change. E) Correct — role, context, examples, format, constraints.

Amazon Rekognition is BEST described as:

1. A managed relational database service
2. A managed vision service for images and video
3. A managed voice synthesis service
4. A managed conversational chatbot builder

Answer: B — A managed vision service for images and video

A) RDS. B) Correct — labels, faces, moderation, PPE, text-in-image. C) Amazon Polly. D) Amazon Lex.

Which SageMaker feature offers pre-built foundation models and solution templates?

1. SageMaker JumpStart with model templates
2. SageMaker Feature Store with online lookups
3. SageMaker Data Wrangler for data prep
4. SageMaker Model Monitor for drift alerts

Answer: A — SageMaker JumpStart with model templates

A) Correct — JumpStart is the model + solution catalog you can deploy in a few clicks. B) Feature Store manages ML features. C) Data prep. D) Endpoint monitoring.

Which AWS capability detects bias in datasets and ML models?

1. Amazon Macie PII discovery
2. AWS Config compliance rules
3. Amazon CloudWatch metrics
4. SageMaker Clarify bias reports
5. AWS CloudTrail audit trail

Answer: D — SageMaker Clarify bias reports

A) PII discovery in S3. B) Resource-config compliance. C) Metrics/logs monitoring. D) Correct — bias detection + explainability (SHAP) for ML. E) Account-level API auditing.

Bias in ML systems can arise from which source?

1. The choice of programming language
2. Historical bias baked into training data
3. The AWS region hosting the training job
4. The number of parallel training instances

Answer: B — Historical bias baked into training data

A) Language-agnostic. B) Correct — biased labels/sampling propagate to the model. C) Region choice is unrelated. D) Compute count is unrelated.

A team wants Bedrock invocation logs (prompt + completion) delivered to S3 for audit. What fits BEST?

1. Enable Bedrock model-invocation logging with an S3 destination
2. Rely solely on CloudTrail management events for the prompts
3. Enable VPC Flow Logs on the Bedrock endpoint interface
4. Turn on GuardDuty ML anomaly detection for the account

Answer: A — Enable Bedrock model-invocation logging with an S3 destination

A) Correct — Bedrock model-invocation logging captures prompts + completions to S3 / CloudWatch. B) CloudTrail logs API calls, not payloads. C) Flow logs capture network metadata, not payloads. D) GuardDuty finds threats, not audit trails.

Where can you obtain AWS audit reports (SOC, PCI, ISO) for compliance reviews?

1. AWS CloudTrail event history
2. Amazon CloudWatch Logs Insights
3. AWS Trusted Advisor cost checks
4. AWS Artifact compliance portal

Answer: D — AWS Artifact compliance portal

A) API audit trail, not audit reports. B) Application logs. C) Cost/security best-practice checks. D) Correct — Artifact is the self-serve portal for AWS audit reports (SOC, PCI, ISO) and customer agreements.

Amazon Bedrock?

Managed service for foundation models. Access multiple models via single API.

What is AWS AI Practitioner (AIF-C01)?

Foundational cert (no prerequisites). Tests AI/ML fundamentals + AWS generative AI services.

Pillars of responsible AI?

Fairness, explainability, privacy, security, safety, controllability, governance, transparency, robustness.

Validation vs test data?

Validation: tune hyperparameters during training. Test: final evaluation, untouched until end.

Bedrock model evaluation?

Compare models on your data. Built-in or custom metrics.

Encryption for AI workloads?

Encrypt data at rest (S3, EBS) and in transit. KMS keys for AI services.