问HN:大型语言模型(LLMs)算不算一种可接受的Lisp?

3作者: tdfirth大约 1 个月前原帖
我这样说是开玩笑,但这是个有趣的想法。 可以很容易地认为一个大型语言模型(LLM)的“程序”是同构的,因为代码和数据都是纯文本。 LLM还提供了丰富的元编程(即编写提示的提示)。 可以说,它们甚至提供了类似于CL的条件系统的功能。 当然,它们并不是在符号表达式上操作,因此称它们为Lisp(或任何其他编程语言)有些牵强,但它们似乎确实共享了许多相同的特性。
查看原文
I say this in jest, but it&#x27;s a fun little idea.<p>It&#x27;s easy to make the case that an LLM &#x27;program&#x27; is homoiconic, as the code and the data are all just plain text.<p>LLMs also offer rich metaprogramming (prompts that write prompts).<p>They even arguably offer features like CL&#x27;s condition system.<p>Of course, they don&#x27;t operate on symbolic expressions, so it&#x27;s a stretch to actually call them a lisp (or any other programming language), but they seem to share a lot of the same properties.