AI-102 Question 34
Select 3You are tasked with designing an Azure AI solution for a healthcare organization that uses machine learning to analyze patient data and predict disease risks. The organization emphasizes the importance of Responsible AI principles, including fairness, accountability, and transparency. Which actions should you prioritize to ensure the solution aligns with Responsible AI principles?
- A
Implement model explainability tools, such as SHAP or LIME, to provide insights into how predictions are made.
- B
Exclude sensitive demographic data, such as age and gender, from the training dataset to avoid bias.
- C
Establish a feedback loop where users can report incorrect predictions or concerns about model performance.
- D
Focus solely on deploying a highly accurate model, as accuracy ensures fairness and eliminates bias.
- E
Conduct fairness tests to measure model performance across different demographic groups.
Show answer and explanation
Correct answers: A, C, E
Explanation
To align with Responsible AI principles, it is essential to prioritize transparency, accountability, and fairness. Model explainability tools provide insights into decision-making processes, while feedback loops enable user accountability and continuous monitoring. Fairness testing ensures equitable outcomes for all demographic groups, addressing potential biases. Simply focusing on accuracy or excluding sensitive data without analysis does not guarantee a Responsible AI solution.
- A. Correct.
Using model explainability tools helps identify how the model arrives at predictions, ensuring transparency and aiding in the detection of potential biases.
- B. Incorrect.
Excluding sensitive demographic data entirely may lead to fairness issues, as it prevents the model from identifying and addressing potential biases related to those attributes.
- C. Correct.
Establishing a feedback loop ensures accountability by allowing users to provide input on the model's performance and flag any issues, facilitating continuous improvement.
- D. Incorrect.
While accuracy is important, it does not inherently ensure fairness. A highly accurate model can still exhibit bias if it performs poorly for certain demographic groups.
- E. Correct.
Conducting fairness tests ensures that the model performs equitably across different demographic groups and aligns with Responsible AI principles of fairness.