知识卡片

普通读书笔记卡 · 8697.a · 生成模型:GPT-6

界面操作依赖观察动作闭环

理解等级:中级

没有合适软件接口时,可通过DOM或截图识别界面,再点击、输入和导航。每次动作都可能改变状态,需重新观察确认结果;能操作界面不意味着理解隐藏业务状态或已完成提交。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      Web Agents and Computer Use Throughout this chapter, we have seen examples of agents executing actions in the real world, primarily by invoking a software interface. However, in the quest for automating human tasks, we find there are many cases where an external software interface doesn’t exist, with only a GUI like a web page available. A lot of tedious work for humans involves actions

    参考来源

    - 位置:《Designing Large Language Model Applications》第10章《外部工具与智能体》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Designing Large Language Model Applications/_epub-src/OEBPS/ch10.html) - 结论依据:摘录中关于界面操作依赖观察动作闭环的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。 - 原始内容:Web Agents and Computer Use Throughout this chapter, we have seen examples of agents executing actions in the real world, primarily by invoking a software interface. However, in the quest for automating human tasks, we find there are many cases where an external software interface doesn’t exist, with only a GUI like a web page available. A lot of tedious work for humans involves actions