Megadose AI progress, ranked and analyzed.

Would You Walk to the Car Wash? Revealing the Salience Bias of Large Language Models in Commonsense Reasoning

· HF Daily Papers ·
The paper says LLMs often know the commonsense answer, but salient distractors can suppress it.

The authors introduce SaliTrap, a benchmark testing how explicit but irrelevant details can hijack commonsense reasoning. Across 12 state-of-the-art models, performance worsened as distractors became denser. A context-free probe recovered over 90% of the failures tied to sycophantic compliance, suggesting the knowledge was present but crowded out. Lightweight inference-time prompting narrowed the gap without retraining. HF Daily Papers' note

score 4

Categories: Research