Scientists at Columbia University have created a robot that can learn without prior programming through deep learning: a breakthrough that is said to be the first step toward self-awareness of machines.
The device consists of an articulated artificial arm and a grab hand like those used in numerous production plants, with New York University scientists giving the ability to "imagine" themselves using a process of self-simulation.
Professor Hod Lipson, director of Creative Machines Lab, said: "If we want robots to become independent, they adapt quickly to unforeseen scenarios by their creators, it is essential that they learn to fake themselves. While the ability of our robot to imagine itself is still rude in comparison to humans, we believe that this ability is on the way to the machine's self-awareness. "
Initially, the robot spent 24 hours behaving like a "babbling baby", moving randomly while trying various tasks. In about a day of intense deep learning, the robot built an internal image of its structure and abilities, with the machine being able to capture objects from specific locations and drop them with 100% accuracy.
Even when relying entirely on his internal self-model – the "imagination" of the machine – the robot was able to complete the pick-and-place task with a success rate of 44%.
A robot that can imagine its own body was created by scientists in a first step towards the self-consciousness of the machine.
PhD student Robert Kwiatkowski, a staff member, said: "It's like trying to get a glass of water with your eyes closed, a difficult process even for humans."
Other tasks included writing text on a board using a marker.
However, to test whether the robot could detect damage to itself, the scientists replaced part of his body with a deformed version. And as a result, the machine was able to recognize the change and bypass it, with little loss of performance.
According to Professor Lipson, self-conscious robots can shed new light on the ancient mystery of consciousness: "Philosophers, psychologists, and cognitive scientists have pondered the nature of self-consciousness for millennia, but have made relatively little progress. We still maintain our lack of understanding with subjective terms as a "reality screen," but robots now force us to translate these vague notions into concrete algorithms and mechanisms. "
Self-conscious robots and computers that are driving or threatening humans have been a rich source of material for novels and science fiction films, and scientists say they are aware of the potential dangers involved in giving robots the gift of self-awareness.
Writing in the newspaper Scientific Robotics, researchers caution: "Self-awareness will lead to more resilient and adaptive systems, but it also implies some loss of control. It is a powerful technology, but it must be treated with care. "
Last October, Pepper, the humanoid robot created by SoftBank Robotics, was the first robot to appear at a UK parliamentary meeting, speaking before the Educational Screening Committee about the future of artificial intelligence in the classroom.