← back
arXivAreeb Gani, Asal Meskin, Gabrielle Kaili-May Liu, Arman CohanTue, Jun 2, 2026, 10:53 AM PDT
score 16.5

Measuring how well AI reasoning models express their confidence

Original: Quantifying Faithful Confidence Expression in Large Reasoning Models

Source: arxiv.org

Writing ELI5 summary…