- Vyacheslav, what are nature-like technologies, what are their tasks?

- The ideologist of the development of nature-like technologies in our country is the president of the Kurchatov Institute Mikhail Kovalchuk. Modern technology is extremely resource intensive. We withdraw resources from nature, but do not return. We use a large amount of fresh water, pollute the environment, burn a huge amount of oxygen. The technosphere that man has created is in conflict with the biosphere.

The energy generation efficiency over the past two hundred years has grown tremendously, but human consumption of energy has grown faster. Over 200 years of its existence, industrial generation has come to the brink of a resource disaster. This is a global challenge because increasing the efficiency of generation is not enough. We need revolutionary changes in the technologies of use, energy consumption.

The idea of ​​nature-friendly technologies is to integrate our industries into natural resources, to become environmentally friendly, energy-efficient, intelligent and practical.

- At the Kurchatov Institute, this direction is called NBIKS technologies, why?

- Nature-like technologies and NBIX technologies are not quite the same thing. The NBICS paradigm, which is an abbreviation, is deciphered as follows: nano- (H), bio- (B), information (I), cognitive (K) and social humanitarian technologies (C). And all these technologies, being combined, converged into a single whole, are an instrument for the development of nature-like technologies. Since the time of Newton, we began to divide nature into separate areas of knowledge. It is time to restore everything again into a single whole. At the junction of sciences, modern and promising technologies for the near future are being born.

- What are the main ideas of NBIKS technologies?

- Nanotechnology is a material science component, they construct materials from individual atoms and molecules. Materials with new properties that did not even exist in nature can become the basis for future technologies. Biotechnologies work with the genome of the cell, with the protein component of all living things, and make it possible to use natural mechanisms for technological purposes. The information component makes all processes intelligent. For example, bio-nanomaterials with a computing device that information technologies give us become intelligent, energy-efficient. With the help of cognitive technologies, we are trying to transfer our knowledge to artificial intelligent systems, providing them with the ability to learn. And all this is interfaced with the socio-humanitarian aspects of technology, since their end user is a person. The meaning of NBICS science is precisely the convergence of knowledge from various disciplines. This is supra-industry knowledge and supra-industry technology.

  • Vyacheslav Dyomin - candidate of physical and mathematical sciences, coordinating director in the field of "Nature-like technologies" SIC "Kurchatov Institute"
  • © scientificrussia.ru

- Has modern science approached the embodiment of these ideas? Or is all this just waiting for us in the future?

- There are quite specific examples of nature-like technologies and NBICS convergence. This is the technology of artificial intelligence and hardware to support it. It is also biotechnology, the production of synthetic organisms, genomic research. In the cognitive area, we influence the psychophysiological sphere of a person, stimulate certain brain regions. The volume of technological solutions and products created using nature-friendly technologies is huge, it amounts to billions of dollars and continues to grow. Without the transition to nature-like technologies, humanity will come to a resource collapse in the foreseeable future.

- There are people and entire countries who deny both global warming and the destructive impact of man on the environment ...

- You know, the opposition point of view can be present in all areas. But the fact that a person most actively affects the environment is obvious. There are problems with the release of greenhouse gases, waste, limited food resources, and especially water. Fuel reserves are being discovered and developed, but with a shortage of fresh water something needs to be addressed. If, for example, India and China with a population of almost three billion people reach the technological level of developed countries, then there will not be enough energy sources, and irreparable damage to the environment will be inflicted. Therefore, we consider nature-friendly technologies to be the only peaceful answer to this global challenge. Otherwise, world wars will begin in the struggle for resources. Separate foci of this struggle we are already seeing.

- Artificial intelligence is developing, it has entered all areas of human life. What ideas are at its core, what can he do now, what is his future?

- Most developed and developing countries have adopted artificial intelligence development strategies. Russia has been developing intelligent technologies for a long time. Expert systems that tried to model high-level human reasoning appeared back in the middle of the last century, when there were no powerful computing resources, good computers. Such systems did not become intelligent, did not think. With their help, you can configure gadgets and other devices, up to microwaves. The current boom in the development of artificial intelligence technologies is associated with a new approach, an attempt to simulate computations at the neural network level, that is, on the principles on which the human brain works.

Trying to restore the architecture of the neural system of biological organisms, we learned how to code the so-called features of objects. Now artificial intelligence algorithms are not programmed, they are trained.

  • Artificial Intelligence Algorithms Learn
  • Gettyimages.ru
  • © John M Lund Photography Inc

- How is artificial neural network training?

- There are two main components of the neural network system: artificial neurons and the connections between them - artificial synapses. Training consists in setting up a huge number of synapses. For example, we have a huge database of images of tables and chairs. And we need to train the neural network to distinguish tables from chairs. We feed thousands of photos to our algorithm. When a neural network is mistaken, a person notes this, submits this error with a special mathematical algorithm back to the network. This is what training is all about. While a person is learning much more efficiently. Two or three examples are enough for him to determine the class of an object, and further training is possible without a teacher. With neural networks, we have not achieved such efficiency, but already roughly know how to do it.

- How soon can artificial intelligence achieve the same efficiency as the human brain?

- It all depends on the task. For example, a neural network does much better with recognizing faces in a crowd. But if recognition occurs among a small number of people, then a person at the same level as her, although here already is losing ground. In many other industries, man surpasses intelligent algorithms by many orders of magnitude. The thing is that artificial neural networks do not understand what they recognize. Scientist John Searle developed a thought experiment - the “Chinese room effect.” A person is in a closed room, communicates with the outside world through a window. He is given a Chinese letter in which he does not understand anything. However, he finds an identical plate in the room, to which several answer options are attached, and gives one of these options to the window. It seems to the outside observer that the person sitting inside knows Chinese. Intelligent algorithms work in a similar fashion today, they are not aware of anything. Therefore, at this level of technology development it is difficult to train.

  • Intelligent algorithms are not aware of anything
  • Gettyimages.ru
  • © Hero Images

“And how do you think, will artificial intelligence soon be able to understand what it is doing?”

- It's hard to predict. But for us it is clear that we need to move towards the so-called pulsed neural networks. Our brain consists of neurons that generate impulses. This is a more energy-efficient system than algorithms based on the now widespread formal, mathematical neural networks. When neurons communicate with each other through impulses, it is possible to establish the principles of their mutual competition in artificial neural networks. Probably, we need fewer parameters to configure the algorithms, but further research in this direction is required.

- What are the dangers of artificial intelligence?

- Already, scientists do not fully understand how the neural network algorithms are being reconfigured. Moreover, there are methods of deceiving them, artificial intelligence makes mistakes. Let's say a person puts on a T-shirt with a print of a rainforest, and the algorithm recognizes it as a palm tree. Adding a certain combination of points to the face interferes with the recognition of a person by artificial intelligence. However, this area, of course, is developing.

- Can there be ethical problems associated with its development?

- The danger of using artificial intelligence is associated with the decision-making process. This applies, for example, to the behavior of unmanned vehicles, where errors in pattern recognition and in decision making can become critical. But the ideas that artificial intelligence will surpass man in everything and try to take the reins of government into their own hands are very far from being realized. While we can not even model the brain of the mouse. This is a daunting task even for modern supercomputers.

- The development of technology affects culture. What do you think, pumping, digitizing the brain, adding some technological elements to the human body, is it still a fantasy or the near future?

- Already there are cochlear devices. These are hearing neuro prostheses that are implanted in the inner ear. There is a retinal neuro prosthesis, which is either installed directly on the bottom of the eyeball or attached in the form of a chip to the primary visual cortex in the occipital region of the brain.

Enormous computing resources are required for the further development of artificial intelligence. In modern gadgets, there can be up to 16 cores in a processor, and for a neuromorphic architecture, hundreds, thousands of cores on a single chip are needed. We develop such systems at the Kurchatov Institute, including on biosimilar principles and biosimilar materials. In particular, synaptic contacts can be implemented as memristors. They are called memorial resistors, resistances with a memory effect. Under the influence of the electrical control signal, such a memristor can change the electrical resistance and transmission efficiency of the signal. A similar process occurs in synaptic contacts between living neurons. A memristor can be scaled to nanoscale sizes, and the density of such elements can be even higher than in the brain. A similar neuromorphic device may appear in the next 5-10 years.

The artificial organs we are working on within the walls of the Kurchatov Institute are now an alternative to the biological organs of donors. In the future, we plan to improve certain human functions. For example, through an implantable device to expand the capabilities of the brain.

See the full version of the interview on the RTD website.

  • Monument to I.V. Kurchatov at the entrance to the Research Center "Kurchatov Institute"
  • © SIC "Kurchatov Institute"