Deep learning uses neural networks to learn tasks that contain one or more hidden layers. What is the nature of deep learning? Is deep learning predictable? More importantly, what are the consequences of deep learning in autonomous machines? The link below, about an experiment at Facebook that took some unexpected turns, is a very interesting article that feeds into perceptions on either the benevolent or malevolent of artificial intelligence (AI). Implementing AI raises questions of whether machine learning should be supervised by humans, partially supervised, or be completely autonomous.