其次，大模型没有天然的执行能力，需要辅以智能体工程来将意图转化为实际操作。工具调用是当前最主流的方式，大模型根据任务需求，生成结构化的函数调用指令，由智能体框架解析后执行相应操作，比如调用天气API、数据库查询、发送邮件等；另一种方式是模拟人类操作，通过视觉识别和模拟操作来"看屏幕、点按钮、填表单"来完成任务，近期大火的豆包手机就是这样完成智能体操作；对于更复杂的任务，智能体还可以配置代码解释器（Code Interpreter / Sandbox），让模型编程运行，这可以极大的扩展智能体的行动边界。
The latest foreboding is from Citrini Research, a little-known US firm that provides insights on “transformative ‘megatrends’”. Its post on Substack, which it called a “scenario, not a prediction”, rattled investors by portraying a near future in which autonomous AI systems – or agents – upend the entire US economy, from jobs to markets and mortgages.
This requirement allows Google to collect intelligence on all Android development activity, including:，更多细节参见同城约会
FT Edit: Access on iOS and web，推荐阅读下载安装 谷歌浏览器 开启极速安全的 上网之旅。获取更多信息
Ghost in the machine。同城约会对此有专业解读
[&:first-child]:overflow-hidden [&:first-child]:max-h-full"