In 2013, a group of researchers wanted to create an AI system that could beat every Atari game. They developed a system called Deep Q Networks (DQN) and less than two years later, it was superhuman. But there was one notable exception. When playing Montezuma’s Revenge, DQN couldn’t score a single point. What was it that made this game so vexingly difficult for AI? Brian Christian investigates.

[TED Ed]