5 of 25 Devan is a machine learning engineer developing an AI-driven loan approval system for a large financial institution. After testing the model, they noticed that applicants from specific neighborhoods are consistently approved at higher rates while others are frequently rejected, even with similar financial profiles. Devan suspects that the training data may reflect historical lending biases and needs to determine if the AI model is making fair lending decisions. What is the most effective way for Devan to evaluate whether the AI model makes fair recommendations? answer Use fairness metrics such as statistical parity and equal opportunity to measure bias in candidate selection Remove applicants' neighborhood data from the model to eliminate geographic bias Compare AI-generated hiring recommendations with human hiring decisions to check consistency Analyze the training data to identify patterns contributing to biased outcomes