SLOW THINKING IMPROVES CONFIDENCE IN LLMS

Summary from the TLDR AI newsletter

Extended chain-of-thought reasoning helps large language models better calibrate their confidence.