Who's Keeping Score? Interactive Steering of LLM-Powered Scoring with Attune
3.80T1 sourcearXiv cs.HC
Source record
Published by arXiv cs.HC (T1 source). The original is at https://arxiv.org/abs/2608.14948.
Pipeline notes
The summary and note below are generated by the signal pipeline — they are Beyond Desk’s reading, not quotations from the source.
SummaryAttune is a mixed-initiative system for LLM-powered text scoring. It uses pairwise comparisons to build global understanding, then derives scoring criteria and rules bottom-up. Users steer scoring logic via examples, direct edits, target distributions, and natural language feedback, validated through technical evaluation and a user study with domain experts in healthcare, law, education, and AI evaluation.
Why it mattersConcrete interface patterns for steering LLM scoring with inspectable, editable criteria - directly applicable to batch text evaluation workflows.
Cited by
No citations on record.
