Robots have come a long way in recent years. No longer are they just simple machines that can do some menial tasks for humans; they have been taught how to walk, feel, and even eat. But now their intelligence has taken an enormous leap forward and robots are now capable of learning new skills on their own.

OpenAI has developed a robot with an AI-powered brain that is able to solve a Rubik’s cube one-handed (something even the most talented of humans have difficulty doing). Dubbed “Dactyl”, this robot was presented with a Rubik’s cube and was able to learn how to solve it with no human help, due to its ability to learn and adapt to its surroundings.

Of course, this is only scratching the surface of what AI-powered robots could be capable of in the future. With the advancement of this technology, we can one day expect robots to be entering every facet of our lives.

In the demonstration video, the Rubik’s cube was placed on Dactyl’s robotic hand, which is then seen to be bumbling its way around it. After many minutes, the robot proceeds to solve the cube. The movement appeared a little awkward and stiff compared to how natural speedcubers perform it with elegance. However, the goal was reached to perfection

Dactyl was shown to undergo many small changes in its index finger, all of which were made possible through machine learning. After the robot figured out how to use its hand to move the cube around, it then curved its back and reached forward toward one of its corners. Once this was achieved, Dactyl was then able to correctly place all seven layers back into the cube.

For robots to be more readily used in our society, it is crucial that they can complete tasks that are difficult for us. Solving a Rubik’s cube one-handed perfectly demonstrates this concept: it is a complex task that most humans find difficult, but it is made simple for robots due to artificial intelligence.

Dactyl robot was implemented with a modern approach in AI called “reinforcement learning”, which ensures that the robot learns and figures its way out on its own of complex situations. This is the goal for the broader AI and robotics community: to develop robots that are capable of performing a variety of functions without being needed to be specially programmed or handled by human beings.

Peter Welinder, a research scientist and robotics lead at OpenAI once said “Plenty of robots can solve Rubik’s cubes very fast. But the important difference between what they did there and what we did here is that those robots are purpose-built. We’re trying to build a general-purpose robot. Similar to how humans and how our human hands can do a lot of things, not just a specific task, we’re trying to build something that is much more general in its scope.”

It is hoped that the Dactyl robot will one day be able to carry out more complex tasks in a similar fashion. Some of these goals may include grasping objects of different sizes, torque, and resistance; some of which are difficult for humans to perform.

The Open AI team hopes to integrate their software with off-the-shelf hardware that will enable the robot to be used on a larger scale for tasks such as logistics. This makes Dactyl an exciting look into the future where robots are afforded a level of intelligence that leads them to be able to conquer any task set before them.