>Just because a model is a better "Google" than my dummy friend doesn't mean that this same friend is more capable at countless cases.

People use LLMs for a lot of things. 'Better Google' is is a tiny slice of that.

>Isn't that just more proof how efficient the human brain is?

Sure. So what ? If a game runs poorly on one hardware and excellently on another, does that mean the game was fundamentally different between the 2 devices ? No, Of course not.