Guidelines for Responsible AI
Human Oversight and Safety
Human oversight is important when AI decisions are high-risk, ambiguous, regulated, or customer-impacting. Safety controls include human review, guardrails, evaluation, monitoring, escalation, and override paths.
Concepts
- Human-in-the-loop review is important for high-risk, ambiguous, or regulated decisions.
- Amazon A2I can route predictions to human reviewers.
- Guardrails, evaluation, monitoring, and escalation paths reduce risk after deployment.
Exam tips
- Amazon Augmented AI (A2I) supports human review workflows for ML predictions.
- Use human-in-the-loop review for uncertain or high-impact outputs.
- Guardrails and monitoring reduce risk but should be paired with escalation paths.