AI-900 Question 48
Select 3A company is developing an AI-powered recruitment system to screen job applications. During testing, they find that the system disproportionately favors male candidates over female candidates for technical roles. What steps should the company take to ensure fairness in their AI solution?
- A
Analyze the training data to check for gender bias and re-train the model if necessary.
- B
Ensure that the AI system is optimized solely for accuracy, as fairness is not relevant in this context.
- C
Apply techniques like re-weighting or re-sampling to address imbalances in the training data.
- D
Conduct regular audits to monitor the system’s predictions for fairness and bias over time.
- E
Avoid introducing any fairness constraints to the model to maintain its performance.
Show answer and explanation
Correct answers: A, C, D
Explanation
Ensuring fairness in AI solutions requires addressing bias in the training data, using techniques to handle imbalances, and monitoring the system over time to maintain fairness. Optimizing solely for accuracy without considering fairness or avoiding fairness constraints can lead to biased and unethical outcomes, particularly in sensitive applications like recruitment.
- A. Correct.
Analyzing the training data for gender bias and re-training the model if necessary is critical to ensure that the system does not learn or perpetuate existing biases.
- B. Incorrect.
Focusing solely on accuracy without considering fairness can lead to biased outcomes, especially in sensitive applications like recruitment.
- C. Correct.
Re-weighting or re-sampling techniques can help address class imbalances or biases in the training data, ensuring the model makes more equitable decisions.
- D. Correct.
Regular audits are essential to monitor the AI system’s behavior over time and ensure that fairness is maintained as new data is introduced.
- E. Incorrect.
Avoiding fairness constraints can result in the system continuing to exhibit bias, which is counterproductive to the goal of building an equitable AI solution.