I see a lot of people eager to play down what LLMs are, using arguments like “LLMs are just … statistical models … that guess words based on … that simulate intelligence”.

I would ask them what exactly do they think humans are?

We found a way of building programs that demonstrate strong linguistic ability with emergent reasoning. Their neural networks are analogous to ours, even if the learning is completely different. Why do they assume that humans are more magical/special than this?
#AI