Confidence Before Answering: A Paradigm Shift for Efficient LLM Uncertainty Estimation
This addresses the need for reliable deployment of LLMs by shifting from answer-first to confidence-first uncertainty estimation, enabling broader downstream applications.
The paper tackles the problem of inefficient uncertainty estimation in large language models by introducing a confidence-first paradigm, where the model outputs its confidence before answering, and demonstrates improved calibration and uncertainty discrimination across math, code, and factual QA benchmarks while preserving answer quality.
Reliable deployment of large language models (LLMs) requires accurate uncertainty estimation. Existing methods are predominantly answer-first, producing confidence only after generating an answer, which measure the correctness of a specific response and limits practical usability. We study a confidence-first paradigm, where the model outputs its confidence before answering, interpreting this score as the model's probability of answering the question correctly under its current policy. We propose CoCA(Co-optimized Confidence and Answers), a GRPO reinforcement learning framework that jointly optimizes confidence calibration and answer accuracy via segmented credit assignment. By assigning separate rewards and group-relative advantages to confidence and answer segments, CoCA enables stable joint optimization and avoids reward hacking. Experiments across math, code, and factual QA benchmarks show improved calibration and uncertainty discrimination while preserving answer quality, thereby enabling a broader range of downstream applications.