For the past several months, I’ve been trying to wrap my head around why AI-writing loves lists. Here is what I mean by lists, as a human might write them: My favorite fruits are concord grapes, ruby red plums, raspberries, nectarines, and mangos…AI writing does something similar—sorta, kinda, almost, somewhat—to this but it doesn’t feel right. As I’ve written before in The Mental Tyranny of AI Writing, these machines can’t make arguments. Or have odd preferences. Because they do word prediction (both forward and backwards, which is a neat technical feature most critics ignore), they make lists in order to triangulate meaning. The problem with lists is they create a relationship between the items in the list. Even if that relationship is vague or just flat-out wrong.
Editors’ Choice: The Curious Question of AI-written Lists: Or, LLMs are Genre Machines
