Understanding "why it works" is one thing, understanding "why it should work this way and not another, and what the alternatives are" is entirely different. AI shows you just one of countless correct implementations. You might understand that single implementation, but not the entire theory behind it