๐ŸŒ€

Hallucination

The phenomenon of AI generating plausible but false content

Hallucination is the phenomenon where LLMs generate factually incorrect information as plausible text. It is divided into Intrinsic Hallucination (content contradicting training data) and Extrinsic Hallucination (fabricating content not in training data). The cause lies in the LLM learning method itself. LLMs are trained to maximize "next token probability," not to "tell the truth." Because they generate the statistically most plausible next word, they cannot distinguish fact from fiction. It can be mitigated with RAG, Grounding, RLHF, etc., but complete resolution remains difficult.

Key Concepts

1

User asks about a specific fact

2

LLM searches for related patterns in training data (probabilistic generation, not exact recall)

3

If related information is insufficient, it "fabricates" statistically plausible content

4

Presents incorrect information with high confidence

5

Risk of user accepting without fact-checking

6

Mitigation methods: RAG (external document reference), Grounding (source citation), uncertainty expression ("it may be...")

Pros

  • Understanding LLM limitations through hallucination research โ†’ safer usage
  • Motivation for advancing hallucination detection/mitigation technology
  • Advancement of factuality benchmarks

Cons

  • Serious problems in high-risk fields like medical/legal
  • Difficult for users to detect hallucinations
  • Cannot be completely eliminated with current technology
  • Reduced LLM trustworthiness

Use Cases

Citing non-existent papers/laws Generating incorrect historical facts Suggesting fictional APIs/functions Incorrect mathematical calculation processes Describing non-existent people/events