Deep learning has captured the imagination of the general public and scientists alike. While this methodology is widely applicable is certainly is not a panacea. It has limitations like all methods, however there are also potential benefits. I will discuss our work on the application of deep learning algorithms to data from the LHC in the contexts of background suppression in the context of traditional-style analyses on ATLAS and work in progress on vision science applications for new physics searches on MoEDAL.