AWS Certified AI Practitioner
Limitations and Risks of Generative AI
Hallucination, inaccuracy, nondeterminism, and interpretability are four different failures with four different fixes. Learn to tell them apart, and see how Amazon Bedrock detects the one you cannot prevent.
Beginner 18 minutes 5 Learning Objectives
- Identify the disadvantages of generative AI named in the exam guide: hallucinations, interpretability, inaccuracy, and nondeterminism
- Distinguish a hallucination from a plain inaccuracy, and both from output variation
- Explain how temperature, top-p, and top-k produce nondeterministic output and what lowering them does
- Describe how the Amazon Bedrock Guardrails contextual grounding check scores grounding and relevance
- Match each failure mode to the mitigation that actually addresses it