Guidelines for Responsible AI Questions

Which statement best describes a transparent or explainable model?

An HR assistant was trained on resumes from a narrow population and begins producing uneven recommendations. What is the best response?

Which responsible AI characteristic refers to a model behaving reliably and predictably when inputs vary or are slightly perturbed?

A model exhibits bias because the training data underrepresents some groups. Which mitigation aligns with AWS responsible AI guidance?

Which responsible AI dimension addresses whether the people who deploy a system can be held answerable for its outcomes?

Which AWS resource provides plain-language documentation describing the intended use cases, limitations, and design choices of an AWS AI service?

Which AWS feature provides human review workflows for ML predictions to support quality, compliance, and accuracy needs?

A high-stakes decision system must combine model predictions with human judgment before action is taken. Which AWS-aligned design pattern best fits?

A medical triage AI must let clinicians override or steer the system in real time. Which responsible AI characteristic does this requirement emphasize...

An organization wants a governance framework that assigns clear ownership for AI incidents, requires documentation of model purpose and limitations, a...

Which AWS tool specifically helps detect bias in ML models and provides feature importance explanations to support fairness and transparency?

Which term describes making AI system behavior and its reasoning understandable and interpretable to stakeholders?

A company wants to monitor a deployed AI model for content that drifts toward unsafe or harmful outputs over time and to maintain governance records o...

A team building a hiring assistant discovers the model gives lower scores to candidates from certain universities. Responsible AI practice requires as...

A team wants to ensure that when their AI system makes high-impact decisions, there is a mechanism for users to question, appeal, or obtain a human re...