Hallucination

When AI confidently produces false information — the #1 failure mode of modern LLMs.

Definition

A hallucination is AI output that sounds plausible but is factually wrong. Caused by next-token prediction without grounding. Detecting hallucinations is a major evaluation category — well-paid because expertise is required to verify domain claims.

Examples

  • Citing a paper that doesn't exist
  • Inventing a fake API method

Related terms

Looking for actual gigs that use Hallucination? Browse all live listings or filter by $100+/hr.