Q16 — AWS AIF-C01 Ch.3

Question 16 of 100 | ← Chapter 3

A company is building a neural network-based model to classify text documents into multiple categories. The company wants to ensure stakeholders can interpret the model’s behavior and predictions. Which method incurs the least operational overhead while satisfying this requirement?

• A. Rely solely on model accuracy and performance metrics
• B. Manually inspect model weights and features to understand internal mechanisms
• C. Develop custom interpretability methods tailored to the selected model architecture
• D. Use model-agnostic interpretability methods, such as Shapley Additive Explanations (SHAP) ✓

Correct Answer: D. Use model-agnostic interpretability methods, such as Shapley Additive Explanations (SHAP)

Explanation

Model-agnostic interpretability methods like SHAP provide intuitive, post-hoc explanations of predictions without requiring deep architectural knowledge or custom development. They apply broadly across model types—including neural networks—and minimize operational overhead compared to manual inspection or architecture-specific solutions, making them ideal for stakeholder communication and compliance needs.