> Is LLM actually a legit proof to show that human brain is just as simple as a bunch of electricity signals flowing around
Does a plane a prove that it's as simple as having wings to fly?
I think it's fairly well accepted that the human brain's cognitive abilities come from its complex electrical network of neurons but I don't think LLMs do much to prove this. We have far better evidence of this from simply observing human brains and the brains of other animals.
That said, the success of LLMs, or ANNs more generally, suggest that the abstractions we have made about the human brain and the nature of cognition are likely correct. Eg, cognitive ability probably doesn't come from the biological matter of the brain, but the way in which that matter is wired up.
> Is the judgement for LLM consistent with human brain's?
It's an interesting point. I've wrote several comments pushing back against people suggesting LLMs are "just predicting the next token" because it seems reasonable to me that for LLMs to be good at next token prediction that they would have to learn how to reason – and there are numerous examples of LLMs appearing to do just that.
I suppose it would be easy to dismiss the human brain in a similar way to how people dismiss LLMs though, "its just some electrical signals flowing around! It doesn't really understand anything".