In October, a researcher at the OpenAI artificial intelligence lab in San Francisco scrambled the tiles of a Rubik’s Cube and placed it in the palm of a robotic hand. Four minutes later, the machine had solved the puzzle. The human world record is only 3.47 seconds, but mastering the cube was a major breakthrough for robotics—the first time a device that works like a human hand had done it. The team had spent months training the robot, even building a computer system that could learn to solve the cube largely on its own. Now they say the experiment’s success indicates that machines could be trained to perform far more difficult tasks. That could one day lead to robots that can reliably sort through packages in a warehouse or to cars that can make more complex decisions on their own. “Solving a Rubik’s Cube is not very useful, but it shows how far we can push these techniques,” says researcher Peter Welinder. “We see this as a path to robots that can handle a wide variety of tasks.”