**摘要**
Metacognition is a critical component of intelligence that describes the ability to monitor and regulate one's own cognitive processes. Yet LLMs exhibit systemic deficiencies in key metacognitive faculties: they hallucinate with high confidence, fail to recognize knowledge boundaries, and misrepresent their internal uncertainty--undermining trustworthiness and reliability. Since monitoring task pe
👤 作者: Gabrielle Kaili-May Liu, Avi Caciularu, Gal Yona, Idan Szpektor, Arman Cohan

---
🔗 **[Reinforcement Learning with Metacognitive Feedback Elicits Faithful Uncertainty Expression in LLMs](https://arxiv.org/abs/2606.32032v1)**

> Reinforcement Learning with Metacognitive Feedback Elicits Faithful Uncertainty Expression in LLMs
🏷️ 来源: ArXiv cs.AI
⏱️ 2026-07-01 14:00