llm quality_error ai_generated true

Factual inconsistency detected in LLM output

ID: llm/hallucination-detection-failure

Also available as: JSON · Markdown
70%Fix Rate
75%Confidence
3Evidence
2023-01-01First Seen

Version Compatibility

VersionStatusIntroducedDeprecatedNotes
any active — — —

Root Cause

LLM generating factually incorrect information.

generic

Workarounds

  1. 88% success Implement RAG with verified sources for factual grounding
  2. 85% success Add citation requirements and verify against knowledge base

Dead Ends

Common approaches that don't work:

  1. Increase temperature for more creative responses 85% fail

    Worsens hallucination

  2. Use longer prompts to force accuracy 72% fail

    More context doesnt prevent hallucination