11How should our assistant behave when it can't or won't help with a request?▼mediumNewAnthropicGoogleMicrosoft◆ premiumHow an assistant fails shapes trust as much as how it succeeds. The interviewer wants different responses for different reasons, each giving users a way forward.Practise this question →Read answer →
19Users say our assistant refuses harmless requests; legal fears it will answer harmful ones. How do you evaluate this?▼mediumNewAnthropicOpenAIMeta◆ premiumTwo teams pulling in opposite directions, each with a real point. The strong answer gives them one shared measurement so the argument becomes a decision.Practise this question →Read answer →