Alignment beyond the verifiable.

LLM judges inherit the fuzzy part of a rubric, so I test models against distributions of human judgments rather than a majority label. Can Reasoning Help LLMs Capture Human Annotator Disagreement? finds that models collapse legitimate disagreement into one overconfident answer, and that RLVR-style reasoning training makes this worse. DIRAS learns calibrated judgments under nuanced natural-language criteria and distills them into small models.

Judgment under fuzzy criteria needs people in the loop, and the tool decides where their judgment goes. Co-DETECT lets an expert refine a rough codebook: it uses annotation uncertainty to surface edge cases the codebook does not cover, induces generalizable rules for them, and leaves the expert in charge of what enters the codebook. AFaCTA calibrates LLM annotators on a small human set and routes items between people and models, so validated labels scale into benchmarks.

The spread of human judgments is irreducible. A model's uncertainty about that spread is not, and it is what a clarification question can resolve. ReProbe reads a frozen LLM's internal states to estimate the credibility of each reasoning step, enabling efficient verification and test-time search without a large process reward model. BELT reads a model's beliefs about its current and future answer from the same hidden states and fits a small stopping policy on them, so one probe serves different accuracy-cost trade-offs and transfers from math to coding and agentic tasks.