Junichi Takeno, a professor at Meiji University in Tokyo (Japan), is one of the leading experts in the field of artificial intelligence, engaged in the creation of machines that work on the same principles as the human brain. His research group has created robots that can recognize themselves in a mirror, recognize emotions, imitate human actions, and make decisions. How far have scientists come close to creating a robot with human consciousness today? Why do you need to teach him to feel heartache? Professor Takeno told about this to Anna Kurskaya, the correspondent of the Social Navigator project of the Russia Today MIA.
- Professor Takeno, what research are you doing now? How far have you and your colleagues succeeded in creating a robot with a human consciousness? - Now my research in the field of creating a thinking robot is aimed at making the robot learn to recognize itself and experience its own feelings using neural networks. My research group has established contacts with the Japanese robotics industry. And the process of assembling the robot with human consciousness has already begun. - How can artificial intelligence be taught to experience emotions, be aware of itself, act in accordance with human values?- My colleagues and I have created a consciousness module based on a double recursive neural network, a kind of analogue of a single human thought. This module, which we call MoNAD (Module of Nerves for Advanced Dynamics), explains most of the features of human consciousness, for example, ten features that were explored by the Austrian philosopher Edmund Husserl. The most important property of a module is its ability to be aware and knows what it is doing.
In addition, we have created a system of consciousness, which is a complex system with layers of MoNAD modules. We call it the MoNAD consciousness system. The system is able to pass the mirror test and be aware of its own states.
The module is able to become aware of its feelings, which can be described based on changes in the stability of the thinking process. Here we relied on the theory of cognitive dissonance, created by the American social psychologist Leon Festinger. In his opinion, if a person's brain is faced with a difficult problem, its explanations and put forward theories will be unstable, internally inconsistent, which leads to the appearance of psychological discomfort.
We imagined that the MoNAD module could express feelings on its own. Our work on building the MoNAD consciousness system allows us to bring robot behavior in line with human values, and I think this will definitely be a new step in research in the field of artificial intelligence.
- To what extent will a companion robot, a communication partner, be in demand in the future? - Communication through the exchange of thoughts between humans and robots is one of the very important aspects of the topic. It is very dangerous to believe that a robot will act for the right reasons. Because there is no evidence that the robot will come from the interests and needs of people. - What is the most difficult task today can artificial intelligence solve? - I think to be creative. - When, according to your forecasts, will be created artificial intelligence, superior to human?- It is very difficult to answer this question, because we still do not know very well what the human mind is. However, when we can model creativity, my research team will be able to implant it in the robot's mind within a few years. This will be a very important event that will allow us to create artificial intelligence that is almost superior to human. “Gradually robots will replace humans in many fields of activity. Can people be expected to indulge in their leisure time while machines create wealth? - This is one of the optimistic scenarios. But the situation is not that simple, I think.- In many books and films, artificial intelligence is destroying humanity. What actions need to be taken today to reduce the risks of such a development of events? “To reduce the risk, scientists must find a way to make artificial intelligence more human. For example, artificial intelligence must feel human pain, including mental pain. - Our conversation is taking place within the framework of the BICA 2019 conference, organized with the participation of NRNU MEPhI. How can you assess the contribution of Russian scientists to the development of the science of artificial intelligence?
- This contribution can be called outstanding. Professor of NRNU MEPhI Alexey Samsonovich and his Russian colleagues have achieved a lot in the development of artificial intelligence. In addition, the great success of the BICA conference is unthinkable without the guidance of Professor Samsonovich, who collaborates with researchers around the world.