#machine learning
Nov 2025
Honesty over Accuracy: Trustworthy Language Models through Reinforced Hesitation
These results establish abstention as a first-class training objective that transforms ``I don't know''from failure into a coordination signal, enabling models to earn trust through calibrated honesty about their limits.
Mohamad Amin Mohamadi, Tian-Hao Wang, Zhiyuan Li
· arXiv.org · 7 citations