AI-102 Question 32
Select 3You are designing an AI solution for a healthcare application that predicts patient diagnoses based on medical records. To ensure compliance with Responsible AI principles, what steps should you take during the planning phase of the solution?
- A
Ensure the model is explainable and provides clear reasoning for its predictions.
- B
Collect and use only the data that is directly relevant and necessary for the model's purpose.
- C
Avoid including any mechanisms for tracking and mitigating biases in the dataset.
- D
Define clear accountability for the AI system's outcomes within your organization.
- E
Prioritize performance metrics such as accuracy over fairness considerations.
Show answer and explanation
Correct answers: A, B, D
Explanation
Responsible AI principles emphasize transparency, fairness, accountability, and ethical use of data. In this healthcare scenario, ensuring explainability, using only necessary data, and defining accountability align with these principles. Ignoring bias mitigation or prioritizing accuracy over fairness would lead to ethical and legal risks in such a sensitive domain.
- A. Correct.
Ensuring explainability is a critical component of Responsible AI, especially in sensitive domains like healthcare, as stakeholders need to understand and trust the model's predictions.
- B. Correct.
Using only relevant and necessary data minimizes privacy risks and aligns with ethical data usage practices, which is a key Responsible AI principle.
- C. Incorrect.
Avoiding bias mitigation directly contradicts Responsible AI principles, which emphasize fairness and the need to identify and reduce biases in AI systems.
- D. Correct.
Defining accountability ensures a clear chain of responsibility for the outcomes of the AI system, fostering trust and ethical governance.
- E. Incorrect.
Prioritizing performance metrics like accuracy over fairness considerations violates Responsible AI principles, which stress the importance of balancing performance with ethical factors like fairness.