I really don't understand how anyone attributes "intelligence" to these automated plagiarism machines.
There are some aspects of this paper that bother me. For example, I think it's absurd to talk about such things as "LLM Reasoning Failures" when there's no reasoning going on at all.
Are we all so conditioned by our education that we think answering questions or writing short essays for an exam is some kind of "reasoning?" It's not.
I'll give an example: Sometimes I meet Evangelical Christian physicians who tell me they don't "believe in" evolution. They might even "believe" that the earth is merely thousands of years old and not billions. They've obviously passed Biology exams to become physicians, they've witnessed the troublesome quirks of the human body that can only be explained by evolution, yet they've never applied any of that to their own internal model of reality. There's an empty space where those models ought to exist. ( Or possibly they are lying to themselves, which is the worst sort of lie. )
With AI it's all empty space. The words go in and the words come out without anything in between.
Whenever I write I'm always concerned that I'm letting the language in my head do my thinking for me; that I'm being the meat based equivalent of an LLM. If I'm doing that I don't really have anything to say. I want all my writing to represent my own internal models of reality as shaped by my own experiences.
LLMs don't have any experiences.
= new reply since forum marked as read