I don't think high-brow / low-brow is a useful framework for understanding these models. They are good at certain things and bad at others and those don't corre

I don't think high-brow / low-brow is a useful framework for understanding these models. They are good at certain things and bad at others and those don't correspond neatly to what a human finds easy and hard.