A simple model is easier to explain, but a complex one is more accurate. What is this?
Why D is the answer
A tradeoff between interpretability and performance
Simpler models are easier to interpret, while complex ones can perform better, so you must trade one against the other.
In detail
- Simple models explain well, and complex ones often perform better.
- Regulation can decide this tradeoff.
Use case
An illustrative scenario, not a real incident.
A bank must explain each loan decision, so it accepts a simpler model at some cost in accuracy.
See it in different ways
flowchart TD
A{"Performance"} -->|"Poor on both"| B["Underfitting: high bias"]
A -->|"Good on training, poor on test"| C["Overfitting: high variance"]
A -->|"Good on both"| D["Good fit"]mindmap
root((Responsible AI))
Bias
SageMaker Clarify
Documentation
SageMaker Model Cards
Safety
Bedrock Guardrails
Transparency
User feedbackflowchart TD
Q{"What does the use case need?"}
Q -->|"Clear explanations"| A0["Simpler model"]
Q -->|"Highest performance"| A1["Complex model"]Key point: Interpretability and performance can pull apart.
Why the other options are wrong
- A. A prompt injection risk in a chat application
- That is a security risk.
- B. A shared responsibility model for AWS customers
- That is about AWS duties.
- C. A limit on the number of tokens per request
- That is a usage limit.