The uniqueness of man lies in ourability to be aware of oneself and represent actions in various situations. We can also learn from mistakes and reflect on what actions are right and which are not. Today, robots are not capable of this. However, things could change in the future. Recently, American researchers have created a robotic arm that is self-aware and able to quickly adapt to changes in the environment. Moreover, the robot works itself, and not according to predetermined instructions. We believe this is the first step in translating consciousness into specific algorithms and mechanisms. The study is published in the journal Science Robotics.
What kind of robots exist today?
Today, most robots are still learning withusing simulators and models provided by people, or through lengthy and painstaking trials by trial and error. Thus, even despite the astounding successes of Boston Dynamics robots, they cannot imitate human behavior and be aware of themselves just like we robots. For now.
By the way, on our channel in Yandex.Zen we often write about the achievements of modern robotics, sign up so you don’t miss anything.
If we want robots to become independent andquickly adapted to various situations and unforeseen scenarios, it is important that they learn to recreate themselves. So says one of the authors of the study, a professor in mechanical engineering, Hod Lipson. Note that since 2007, Israeli and American engineers have been creating robots that learn not only to fly, swim and walk, but also to laugh and cry.
It is interesting: Humanoid robots become reality
During the study, Lipson and his colleague RobertKwiatkowski used a robotic arm. Initially, the robot moved randomly and made about a thousand movements in different directions. Then the robot applied deep learning, a modern machine learning technique, to create images of models like himself. The first "car models" were inaccurate and the robot did not know what they were and how they were connected. But after less than 35 hours of training, the painted model began to correspond to the physical structure of the robot. Researchers note that the robot did not receive any instructions for modeling and assembling models. In other words, he learned everything himself and understood what he looked like.
The danger of robots
Lipson notes that self-awareness is the keyin order to allow robots to move away from the limitations of the so-called "narrow artificial intelligence", which can only be aimed at one goal. According to the researchers, what a newborn baby does in a crib — how he learns the world around him — is perhaps an advantage of the evolutionary origin of human self-awareness. Although the ability of a robotic arm to be aware of itself is not yet comparable to human, scientists believe that this is the beginning of the era of self-awareness of machines.
Philosophers, psychologists and scientists throughoutFor thousands of years, they have been reflecting on the nature of self-awareness, but have made relatively little progress. We still hide our misunderstanding with subjective terms, but today robots are forcing us to translate these vague concepts into specific algorithms and mechanisms. Lipson and Kwiatkowski claim to be well aware of the ethical implications of creating a self-conscious machine.
Read more news from the world of amazing scientific discoveries on our Telegram news channel.
According to Lipson, self-awareness will lead tothe creation of more flexible and adaptive systems, but this also implies some loss of control over the situation. This is a powerful technology, but it must be handled with care. Now scientists are working to ensure that robots recreate not only their body, but also their own mind. We hope they know what they are doing.