Question

A healthcare company deploys a Bedrock-based patient-facing assistant. Leadership wants to add aggressive content filters that block any response touching on medication dosages, symptoms, or diagnoses to eliminate liability risk. The product team warns that this will make the assistant refuse most legitimate questions and frustrate users. Which statement best describes the responsible-AI trade-off the team is facing?

Accepted Answer

Increasing safety restrictions can reduce the assistant's usefulness, requiring a balance between protective guardrails and helpful functionality

Answer

Adding guardrails always improves both safety and helpfulness simultaneously with no downside

Answer

Content filters improve model explainability, so more filtering makes the model more transparent

Answer

Guardrails eliminate bias in the training data, so aggressive filtering removes the need for fairness testing