In the past few years, methods of artificial intelligence and in particular deep learning (DL) have been broadly discussed in the context of nanophotonics and manifold formerly impossible applications, were unexpectedly enabled thanks to DL. Probably the most extensively discussed application of deep artificial neural networks (ANNs) in nanophotonics is the inverse design of photonic devices and nanostructures. But an abundance of further applications exits, ranging from enhanced and robust data‐processing and “knowledge discovery”, over the ultrarapid surrogate modeling for the acceleration of physics‐simulations, to efficiently solving experimental inverse problems like sparse data reconstruction. In this chapter, we will discuss selected pioneering applications which are enabled thanks to DL‐based techniques. We provide a critical review of some of the aforementioned novel possibilities, and we discuss in particular possible pitfalls and weaknesses that are inherent to the data‐driven nature of numerical methods based on ANNs.