如何将大型语言模型（LLMs）用作教师？
我看到许多评论对教育系统的衰落表示遗憾，原因是学生将所有作业都交给大型语言模型（LLMs）。在这些讨论中，人们声称LLMs可以成为有效的教师，但学生们并没有正确使用它们。根据我的经验，无法判断一个主题中真正重要的内容，以及被迫采用“助手”模式的默认节奏，使得使用LLMs的效果大致与直接阅读维基百科页面相当，甚至更差。那么，人们是如何做到的呢？如果你能够直接从LLMs中学习新知识，你是怎么做到的？
查看原文
I’ve seen many comments lamenting the fall of the education system due to students handing off all of their work to LLMs. In those threads, people claim that LLMs _can_ be effective teachers but that students just aren’t using them right. In my experience, the inability to tell what’s actually important about a topic and the forced “assistant” default cadence makes it roughly on-par or worse than just reading the Wikipedia page. So how are people doing it? If you are able to learn new things directly from LLMs, how are you doing it?