Post

Demonstrating reasoning and language capabilities in computers alters our understanding of self, even if we don't particularly like the alteration

This is the crux of the debate, right? Whether it's doing mimicry or reasoning. I struggle to see how you could view it as purely mimicry. It has to tie together multiple novel thoughts in a logical construction to create its responses.

This perhaps says something about the logic inherent in language, or in the surprising observation that compression has an equivalence to intelligence due to the nature of what prediction means, but here we are: the LLMs work through novel questions and scenarios regularly

That's a meaningless statement. People task AIs with working through scenarios and answering questions. Some of those problems are incredibly complex but at no point are we seeing anything but machine accomplishing a given task. There's no motivation or independent reasoning .

I mean yeah but that *in itself* is weird in a way that changes how we think about ourselves, right? That a language prediction system with no motivation or independent reasoning per se is capable of all of this.

If you told someone a decade ago that in ten years giant statistical models of human language would be capable of hacking computers with very little in the way of human direction, people would ask if you were high.

People have been thinking about it and building towards it, but they *did not initially expect to get to this level of capability this fast.*

Okay, what level of capability do you think I attribute to LLMs? Because I don't think I've made any claims about their abilities that aren't just... citing things that literally happened, so I'm not sure how I'm overstating.

0 reposts13 likes

Yes, but I can't read your mind. I'm trying to understand what you think I think, because I'm not sure if it's the same as what I actually think.