📄 中文摘要
AI 代理的记忆容量增加可能导致其效率降低。随着交互日志的积累，这些日志变得庞大，充满了无关内容，使用起来愈加困难。更多的记忆意味着代理需要在更大体量的过往交互中搜索，以找到与当前任务相关的信息。缺乏结构的记录使得这些信息混杂在一起，难以提取出有用的知识。因此，重新思考 AI 代理的记忆管理方式显得尤为重要，以提高其在实际应用中的有效性。
从原始交互到可重用知识：重新思考 AI 代理的记忆
📄 English Summary
From raw interaction to reusable knowledge: Rethinking memory for AI agents
Increasing the memory capacity of AI agents can paradoxically reduce their effectiveness. As interaction logs accumulate, they become large, filled with irrelevant content, and increasingly difficult to navigate. More memory requires agents to sift through larger volumes of past interactions to locate information pertinent to the current task. Without proper structure, these records become a chaotic mix, complicating the extraction of useful knowledge. Rethinking the management of memory for AI agents is crucial for enhancing their effectiveness in practical applications.