Algorithm studies images to produce a video of what might happen next

Given a still image, the deep-learning system from Massachusetts Institute of Technology generates videos that predict what will happen next in a scene.