Responsible AI as an Engineering Practice

1. What is the primary distinction between a determinative AI system and an advisory AI system?

Determinative systems require human approval for all decisions Advisory systems provide recommendations that inform human decision-making Determinative systems can be tweaked without deployment Advisory systems have more complex architecture

2. What is a recommended practice for maintaining auditability in AI outputs?

Using random data for testing Implementing retrieval-augmented generation with citations Avoiding human input in decision-making Focusing solely on algorithm efficiency

3. Which of the following is not considered a part of the control plane for responsible AI?

Feature toggles Role-based access Model optimization Audit logging

4. Why should AI systems use deterministic logic wherever possible?

It makes systems easier to explain It requires less computing power It avoids human intervention altogether It eliminates the need for monitoring

Continue: MLOps Vocabulary for People Who Already Do It → Review this lesson Retake quiz