Diyalog Tamamlama (Yapay Zekâ ve Etik)Soru #10777
Soru sayfası
Liam: Our machine learning model performs with ninety-five percent accuracy on credit risk scoring.
Sophie: That is impressive, but have you audited the feature weights to see if historical biases are being reinforced?
Liam: ----
Sophie: Exactly. If we don't verify that, we risk automating discriminatory lending practices under the guise of objective mathematics.
Sophie: That is impressive, but have you audited the feature weights to see if historical biases are being reinforced?
Liam: ----
Sophie: Exactly. If we don't verify that, we risk automating discriminatory lending practices under the guise of objective mathematics.
- A)Mathematics can never contain errors or biases regardless of how training data is collected.
- B)We already decommissioned all company servers and decided to close the financial firm today.
- C)Not yet, and that is a genuine concern because training on historical loan approvals might inadvertently penalize applicants from certain zip codes.
- D)Why would anyone care about credit scores when evaluating loan repayment probabilities?
- E)I would prefer that we switch to using dice rolls instead of algorithmic evaluation.
Cevabı göster
Doğru cevap: C) Not yet, and that is a genuine concern because training on historical loan approvals might inadvertently penalize applicants from certain zip codes.