Deep learning is famously biologically inspired and many of the major concepts in deep learning are intuitive and grounded in reality. The fundamental truth of deep learning is that it’s hierarchical - the layers in a network and the representations they learn build on each other. This is also the case with reality: electrons, protons, neutrons -> atoms -> molecules -> … It makes sense that the best way to model a hierarchical world is hierarchically, and this is why deep learning has been so successful in providing simple, elegant, and general solutions to very difficult problems.
The biggest mistake that can be made in artificial intelligence is to try to judge an algorithm by its results the moment we get it, without taking into account the progress that can be made by using more and better data. Criticizing Amazon’s Echo by saying that it is little more than a glorified radio clock overlooks the fundamental thing: that with eight million devices on the market, Amazon’s chances of improving Echo’s intelligence are practically unlimited, and that means that over time we will understand things better, gradually reducing its errors, meaning that it will soon be a device that we will end up wondering how we ever lived without it.