From a data appropriateness and bias perspective, which of the following should be of GREATEST concern when reviewing an AI model used in a credit scoring system?

Question

Options

• AThe model incorporates the applicant's loan history to assess spending habits.
• BThe model utilizes historical credit data to predict future credit behavior.
• CThe model considers the applicant's income level as a key factor in the credit decision.
• DThe model uses postal codes as a primary factor in determining creditworthiness.

• A
  4% (1)
• B
  12% (3)
• C
  8% (2)
• D
  77% (20)

Explanation

Using postal codes as a primary factor in credit decisions is the greatest concern because postal codes can serve as a proxy for race or ethnicity due to historically segregated residential patterns - a practice historically known as 'redlining.' Even though the model does not explicitly use race, the outcome can be discriminatory, which is a form of indirect or proxy bias. This violates fairness principles and may breach fair lending regulations (e.g., the Equal Credit Opportunity Act). The other factors - loan history (A), historical credit data (B), and income level (C) - are legitimate and widely accepted inputs for credit scoring that have a direct causal relationship to creditworthiness.

No community discussion yet for this question.