我们能否，以及是否应该赋予大语言模型一丝“异想天开”？
1 分•作者: watmough•8 个月前
我周末花了很多时间，真正掌握了使用我的 claude.ai Max 计划的技巧，使得 claude 能够以非常符合我个人喜好和倾向的方式创建系统和文档。<p>而且，作为一个患有自闭症孩子的父亲，我自己可能也有点自闭症倾向，我与孩子们交流时会用到很多荒诞的幽默。<p>有没有人认真思考过，我们是否能让 LLM 变得像维克多·刘易斯-史密斯或类似的讽刺作家那样既幽默又令人不安？
查看原文
I spent the weekend really nailing down my usage of my claude.ai Max plan such that claude creates systems and documentation in a way strongly aligning with my own tastes and inclinations.<p>And as a father of one quite autistic child, and likely being somewhat on the spectrum myself, my communication with the children involves a lot of absurd humor.<p>Has anyone thought really hard about whether we can make LLMs as funny and disturbing as, say, a Victor Lewis-Smith, or similar satirists?