why each new mind-blowing advance in deep learning isn't actually impressive

Honest question: can anyone cite some examples of AI living up to the hype?

(Meaning people actually using AI for impressive, impactful projects, not just "mind-blowing" tech demos.)