研究揭示AI智能体技能提升机制及检索瓶颈
原标题：Study explains why AI agents benefit from "skills" and when they fail
AI 摘要
普林斯顿大学和加州大学圣迭戈分校的研究人员通过超过8000次测试，研究了技能（skills）如何提升AI智能体的性能。研究发现，技能主要通过提供程序性指导（procedural grounding）帮助智能体，而非补充事实知识，这一机制在约66%的案例中有效。然而，当技能库从5个扩展到100个条目时，检索命中率从29.6%降至3.3%，表明技能检索是主要瓶颈。研究者建议将技能使用视为生命周期，并改进技能的创建、检索和应用方法。
正文节选
Study explains why AI agents benefit from "skills" and when they fail Key Points - Researchers at Princeton University and UC San Diego ran more than 8,000 test runs to study how "skills" improve AI agents. These skills work as compact instructions for specific tasks. - The results show that skills help less through factual knowledge than through set procedures. This procedural grounding boosted performance in nearly 66 percent of cases. - One weak spot remains finding the right instructions. Wh