Why Models Hallucinate Instead of Admitting Ignorance: Benchmarks Reward Correctness, Not Honesty
ziv_ravid · x · 2026-07-04
The authors of the "Mirage" research explain why models tend to hallucinate rather than admit "I can't tell": it's shaped by evaluation mechanisms. Existing benchmarks only reward correct answers, not whether they are grounded. Thus, guessing based on priors has a positive expected return, while skipping yields zero—making hallucination the optimal strategy under current evaluation systems.
More from Research
- Google DeepMind pilots world's first double-blind evaluations for frontier models — scychan_brains · 2026-10-26
- Rewriting scaling law study: optimal synthetic data share rises from 10% to 32% with compute — heghbalz · 2026-10-26
- FLASH: GPU-native deformable sim learns cloth folding in minutes on one RTX 5090 — ChongZzZhang · 2026-10-26
- LEMON: zero-shot LLM misbehavior monitor flags hallucinations from attention maps — yoavartzi · 2026-10-26
- Behind the EMNLP keynote: LMs as simulated learners to generate child language acquisition hypotheses — najoungkim · 2026-10-26
- The paper behind the EMNLP 2026 keynote on bigram 'simulated learners' — najoungkim · 2026-10-26