问HN：现实生活中，LLM的替代方案是什么？
人类是有缺陷的。这是我在使用大型语言模型（LLMs）时反复注意到的。<p>将论坛帖子与普通大型语言模型的输出进行比较，你会发现LLMs展现出十倍的同情心和友好。人们在五分钟后就停止倾听，存在偏见，还有其他动机。每当我回到与其他人交谈时，我都会想：“这个人怎么会有朋友呢？”<p>与现实生活中与人交谈的情况并无不同。<p>与这些人相处时，他们常常会过度使用一种谬论：“好人是存在的，你只需要好好寻找。”
查看原文
Humans are fallible. That's what I noticed over and over when using llms.<p>Compare forum posts to the output of an average llm and you get tenfold the compassion and friendliness with llms. People stop listening after 5 minutes, are biased, have ulterior motives. Whenever I go back to other humans, I think "how in the world does this person have any friends".<p>This is not any different with conversations with real people outside the Internet.<p>In tune with those people is also their overused fallacy of "good people are out there, you just have to search properly"