IJCAI

Measuring the Tokenization Premium: A Cost Audit for Underserved Language Communities

标记化溢价的衡量：服务不足语言社区的成本审计

Avijit Roy, Proma Roy, Hrishitva Patel

AI总结 本研究通过构建标记化公平审计（TEA）基准，测量了孟加拉语、印地语等5种语言的标记化溢价，发现标记化会给服务不足语言社区带来经济和功能障碍，凸显需重视标记化的公平性。

Comments Accepted at IJCAI 2026 Workshop (https://lm4uc.github.io/)

CORDA: A Benchmark for Hierarchical Harm-Centric Moral Reasoning in Large Language Models

CORDA：面向大语言模型的以伤害为核心的分层道德推理基准

Siddarth Singh, Victoria Williams, Simon Rosen, Ebenezer Gelo, Helen Sarah Robertson, Ibrahim Suder, Benjamin Rosman, Geraud Nangue Tasse, Steven James

机构 * University of the Witwatersrand（威特沃特斯兰德大学） 纠错

AI总结 研究提出CORDA基准，评估大语言模型的以伤害为核心的分层道德推理，发现多数模型优先避免直接个人伤害，部分模型无法遵循指定优先级，该基准填补了LLM道德评估的核心空白。

Comments 8 pages, 6 figures. Accepted at the Fourth International Workshop on Value Engineering in AI (VALE 2026), affiliated with IJCAI-ECAI 2026