AI-900 Question 46
Select 3A company is building an AI-powered hiring system to screen job applicants. During testing, the team notices that the system disproportionately rejects candidates from certain demographic groups. What steps should the team take to ensure fairness in the AI solution?
- A
Analyze the training data for bias and ensure it represents diverse demographics.
- B
Exclude demographic data from the training dataset to avoid potential bias.
- C
Regularly test the AI model for unequal impact across different demographic groups.
- D
Trust the AI system's decisions as it was trained on a large dataset.
- E
Incorporate human review into the decision-making process to mitigate bias.
Show answer and explanation
Correct answers: A, C, E
Explanation
Fairness in AI solutions requires a proactive approach to identify and address potential sources of bias. This includes analyzing training data, regularly testing model behavior for fairness, and incorporating human oversight to ensure the system does not unfairly disadvantage any group. Excluding demographic data or trusting the AI system without scrutiny are insufficient strategies for ensuring fairness.
- A. Correct.
Analyzing the training data for bias and ensuring diverse representation is critical to building a fair AI system. Biased training data can propagate or amplify inequities.
- B. Incorrect.
While excluding demographic data may seem like a way to avoid bias, it can actually prevent the model from identifying and correcting existing biases in the dataset.
- C. Correct.
Regularly testing the AI model for unequal impact is essential to ensure that the system performs fairly across different demographic groups.
- D. Incorrect.
Blindly trusting the AI system is not a valid strategy for fairness. Even large datasets can contain hidden biases that need to be addressed.
- E. Correct.
Incorporating human review helps identify and mitigate potential biases that the AI system might inadvertently introduce.