Study on the Cognition–Action Disconnect in Tool-Using Agents

工具使用代理认知与行动脱节机制研究

An interpretability paper studies tool-using agents and finds models often recognize when to call a tool but fail to act, with a cognition-to-action mismatch rate of 26%–54%.

Why it matters: HKR-H/K/R all pass: the story has a sharp agent-failure hook, a 26%-54% mismatch rate, and clear relevance to tool-use reliability. Source detail is thin, with paper name, models, and task setup not disclosed.