Abstract: Advances in deep reinforcement learning have made it possible to create artificial intelligence-based agents for games that use visual information to make decisions as accurately as humans.