GCP Generative AI Leader Practice Question 90: Explainability and Accountability
Google Cloud Generative AI Leader · Test 9 · question 10 of 10
A lender is deploying a gen AI assistant that helps decide whether to approve loan applications. Regulators require the company to be able to show which factors contributed to a specific denial, and internal governance requires a named owner within the company who answers for the system's outcomes and can order it changed or taken offline. Which pair of responsible AI principles does this requirement most directly address?
Show answer and explanation
Answer:
Explainability is the ability to describe why a model produced a specific automated decision in terms a human can understand, which is what regulators are asking for when they want to know which factors led to a denial. Accountability means an organization owns responsibility for the system's effects, including a governance structure with a named owner who can act on the outcome. Together those two principles match both parts of the requirement, so option D is correct.
- A is wrong because privacy and security protect personal data from misuse or exposure; they do not explain why a particular decision was reached or who owns the outcome. - B is wrong because fairness addresses equitable outcomes across groups and robustness addresses stability under varied inputs, neither of which is what is being asked for here. - C is wrong because transparency about AI use and safety against harmful output are real principles, but the scenario is specifically about explaining a decision's factors and naming an accountable owner, not about disclosure or harm prevention.
Reference: Accountability