Guide-linked quiz · Medium Level
AI Credit Limit Management Quiz
Explains AI-assisted credit-limit decisions, accurate adverse-action reasons and controls for monitoring model recommendations.
Question 1 of 8
What does an AI credit-limit recommendation represent?
Preview every question8 questions
Answers and explanations appear as you take the quiz.
- What does an AI credit-limit recommendation represent?
- Evidence for a decision process, not a borrower-specific fact
- A final policy decision whenever model confidence is high
- A definite prediction of this borrower’s future repayment
- A reason to bypass normal account-review controls
- Under current Regulation B, which credit-limit decision is an adverse action?
- Declining a purchase that exceeds an unchanged existing limit
- Changing only an internal risk score without acting on the account
- Refusing an applicant’s request to increase available credit
- Sending a balance update without changing available credit
- What must a covered Regulation B adverse-action reason statement convey when an algorithm informs the decision?
- Only the model’s composite score, without principal factors
- Specific principal reasons tied to factors actually considered
- Every model feature, even if it did not influence the action
- A generic statement that internal policy was not met
- Why can training on historical credit-limit decisions be risky?
- Past decisions are always random
- Data cannot contain policy choices
- Models never use spending history
- The model may reproduce old policies or access patterns
- What should a lender distinguish when designing a limit workflow?
- Only credit scores and income
- Only online and branch accounts
- Increasing, decreasing, freezing and reviewing a limit
- No distinctions are needed
- What should a reason-code mapping verify?
- That stated reasons match factors actually used
- That codes sound customer-friendly
- That every notice uses the same reason
- That no model documentation is retained
- Which outcomes should a lender monitor for its limit model?
- Only retrospective model accuracy, without customer outcomes
- Only the average size of approved limit increases
- Only approval rate, without repayment or correction data
- Errors, stability and downstream customer outcomes
- What should happen when a customer disputes inaccurate input data?
- Ignore it if the model score is high
- Provide a correction or review path
- Replace the model with a marketing rule
- Delete all decision records
Keep testing yourself
More quizzes picked for your level.