AIF-C01 Practice Question 18: Model Documentation Transparency
AWS Certified AI Practitioner (AIF-C01) · Test 2 · question 8 of 10
An insurer's model risk committee requires a single, versioned record for every ML model before it goes to production. The record must show the model's intended uses and out-of-scope uses, its risk rating, training details, and evaluation results, and it must be exportable for auditors. Which AWS capability is designed for this?
Show answer and explanation
Answer:
Amazon SageMaker Model Cards document the critical details of an ML model in one place for governance and reporting. That includes its intended uses (and uses it isn't meant for), a risk rating of unknown, low, medium, or high, training details and metrics, and evaluation results. Every edit creates a new version, which gives an immutable history, and a card can be exported to PDF for stakeholders such as auditors. This kind of documentation is a core tool for transparency in responsible AI (option A).
- B is wrong because AWS Artifact provides AWS's own compliance reports and agreements, such as SOC reports. It doesn't document your models. - C is wrong because CloudTrail logs API activity in your account. That's useful audit evidence, but it doesn't describe a model's purpose, risk, or evaluation. - D is wrong because Feature Store stores and serves ML features for training and inference. It isn't a model documentation record.
Reference: Amazon SageMaker Model Cards