AI-900 exam dumps

AI-900 practice question 47 of 286

Microsoft Azure AI Fundamentals. Free level, Microsoft. Free question with the correct answer and a full explanation.

AI-900 Question 47

Select 3

A company is building an AI-powered recruitment system to screen job applications. During testing, the team notices that the system tends to favor candidates from a particular demographic group. Which considerations should the team focus on to ensure fairness in their AI solution?

  1. A

    Ensure the training data is diverse and representative of all demographic groups.

  2. B

    Clearly define fairness metrics appropriate to the recruitment context.

  3. C

    Prioritize the accuracy of the model above all other considerations.

  4. D

    Continuously monitor the system for bias even after deployment.

  5. E

    Remove all demographic information from the training data to avoid bias.

Show answer and explanation

Correct answers: A, B, D

Explanation

To ensure fairness in an AI solution, especially in sensitive areas like recruitment, the team must focus on representative training data, appropriate fairness metrics, and ongoing monitoring to address existing or emerging biases. Removing demographic information alone is insufficient, as biases can still arise through indirect correlations.

  • A. Correct.

    Ensuring the training data is diverse and representative reduces the risk of bias and helps the model generalize across all demographic groups.

  • B. Correct.

    Defining fairness metrics specific to the context of recruitment allows the team to measure and evaluate fairness effectively.

  • C. Incorrect.

    While accuracy is important, prioritizing it above fairness could lead to biased outcomes, especially in sensitive applications like recruitment.

  • D. Correct.

    Continuous monitoring is essential because biases can emerge over time as the system interacts with new data or changing environments.

  • E. Incorrect.

    Simply removing demographic information may lead to unintentional proxy biases, where other features act as stand-ins for the excluded demographic data.

Timed practice exam

Take a AI-900 practice test under exam conditions

60 questions in 30 minutes, drawn from this bank, with a score report and a per-question review when you finish.

Start timed exam