AOR-Bench: Do Large Audio Language Models Over-Refuse Pseudo-Harmful Queries?
A benchmark for over-refusal in large audio language models — how often they decline pseudo-harmful but benign spoken queries, and what drives it.
jiaxi-yang
•
•
1 min read
Position: Breaking the Dual Curse of Multilingual AI Requires Socio-Technical Guardrails, Not Post-Hoc Alignment
A position paper arguing that the dual curse of multilingual AI — 35% harmful generation and near-random reward-model accuracy in low-resource languages — cannot be fixed by …
