AIF-C01 Question #109: Real Exam Question with Answer & Explanation

Question

An ML research team develops custom ML models. The model artifacts are shared with other teams for integration into products and services. The ML team retains the model training code and data. The ML team wants to build a mechanism that the ML team can use to audit models. Which solution should the ML team use when publishing the custom ML models?

Options

• ACreate documents with the relevant information. Store the documents in Amazon S3.
• BUse AWS AI Service Cards for transparency and understanding models.
• CCreate Amazon SageMaker Model Cards with intended uses and training and inference details.
• DCreate model training scripts. Commit the model training scripts to a Git repository.

Explanation

Explanation

Amazon SageMaker Model Cards are purpose-built for exactly this use case - they provide a structured, standardized way to document model details such as intended uses, training data descriptions, evaluation results, ethical considerations, and inference configurations, making them ideal for auditing custom ML models within an organization. Model Cards are natively integrated with SageMaker, allowing the ML team to attach documentation directly to model artifacts and share them systematically with other teams.

Why the distractors are wrong:

• Option A (S3 documents) is too unstructured and ad hoc - there's no standardized format for auditing, making it difficult to ensure consistency across models.
• Option B (AWS AI Service Cards) are transparency documents published by AWS itself about its pre-built AI services (like Rekognition or Comprehend), not a tool for customers to document their own custom models.
• Option D (Git repository for training scripts) only captures code, not the broader model context needed for auditing (e.g., intended use, bias metrics, evaluation results).

Memory Tip

Think: "Cards are for custom, AI Service Cards are for AWS." When a team builds their own models and needs structured audit documentation, SageMaker Model Cards is the answer. AWS AI Service Cards are AWS's own documentation - not yours to create.

No community discussion yet for this question.