Question 15
Domain 1: Plan and Manage an Azure AI SolutionYour team is considering using Azure AI services to analyze HR-related documents. The legal team requires that before any AI system makes decisions about employees, there must be documented human oversight and an audit trail. Which Responsible AI components should you implement?
Correct answer: B
Explanation
Human oversight is required because the system must not make employee decisions without review, so "Human-in-the-Loop" reviews provide documented human oversight. Audit logs create the required audit trail by recording decisions and actions, and documenting the AI governance framework shows how responsibility, review, and accountability are enforced.
Why each option is right or wrong
A. Fine-tune the model on HR policies only
B. Implement Human-in-the-Loop reviews, maintain decision audit logs, and document the AI governance framework
The requirement maps to the governance and accountability controls in the NIST AI Risk Management Framework 1.0, especially the Govern function, which calls for documented oversight, traceability, and accountability for AI decisions. In practice, Human-in-the-Loop review provides the required human oversight before employment-impacting decisions are finalized, while decision audit logs preserve a traceable record of inputs, outputs, and reviewer actions for later inspection; documenting the governance framework shows the control structure and responsibilities that auditors expect to see.
C. Use Managed Identity so no API keys are stored
D. Enable content filters to prevent inappropriate outputs