Bài kiểm tra có liên kết hướng dẫn · Trung bình Cấp độ
Bài kiểm tra quản lý hạn mức tín dụng AI
Giải thích các quyết định giới hạn tín dụng được AI hỗ trợ, lý do hành động bất lợi chính xác và các biện pháp kiểm soát để giám sát các đề xuất mô hình.
Câu hỏi 1 của 8
What does an AI credit-limit recommendation represent?
Xem trước mọi câu hỏi8 câu hỏi
Câu trả lời và giải thích xuất hiện khi bạn làm bài kiểm tra.
- 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