Deep neural network models have substantial advantages over traditional and machine learning methods that make this class of models particularly promising for adoption by actuaries. In the past few years, many different applications of these models have appeared in the actuarial literature. Drawing on recent work, in this talk we will journey through recent advances in deep learning applied to actuarial topics, covering advances in representation learning for actuarial purposes, structuring deep networks for explainability and uncertainty estimation. We also cover advances in model interpretability and avoiding direct and indirect discrimination in supervised learning and will conclude with open research topics in AI applied within Actuarial Science.