QABattle

AI & LLM Evals / Quiz

Hallucination Detection Methods

Confident and wrong is the worst failure an LLM has. These are the concrete methods for catching it before a user does.

Difficulty
Easy
Format
Quiz
Points
100
Estimate
9 min

// MISSION BRIEF

Your Mission

"Hallucination" is not one thing, and neither is detecting it. Grounded (RAG) hallucinations, closed-book fabrications, and unsupported citations each have detection methods. This quiz covers the toolkit.

Scoring: equal weight per question.

// FIRST CONTACT

Battle teaser

For a RAG answer, the most direct hallucination check is:

  1. AFaithfulness/groundedness, does every claim trace to the retrieved context?
  2. BAnswer length
  3. CResponse latency
  4. DThe user's rating
Answers, scoring, hints, and the full battle stay sealed.

// SKILL TAGS

ai-evalshallucinationfaithfulnessnliself-consistency