Back to feed
arXiv cs.AI·

Knowing When to Ask: Self-Gated Clarification for Hierarchical Language Agents

Signal
72
Hype
18
In three linesACTION-RATING embeds clarification requests directly in the agent's action space, competing with navigation at every decision point. On Harmonized Tariff Schedule classification (30k-node taxonomy, 9 LLMs), two information-seeking modes emerge: mandatory (no viable branch) and opportunistic (residual uncertainty). Information-Seeking Effectiveness rises from 50% to 74%.
Read source
Your take?
AI AgentsReasoningEvals

Summary generated by Claude — human-verified