I would agree on that in case of ANN, maybe even in case of CNN and text processing networks. However, when we talk about GANN, PINNs and more complicated networks, constrains, generative capabilities and other aspects does make it DEEP LEARNING and not just memorizing.