Section introduction

The article discusses the task of creation of Artificial General Intelligence (AGI), that is, an artificial intelligence system that approaches the functional capabilities of natural intelligence. Emphasis is made on the leading role of Sberbank in updating and organizing a special research program in order to develop this problem, which is of strategic importance for Russia. It has been established that the successful implementation of this program presupposes solving fundamental methodological issues that require input of philosophers - specialists in the field of epistemology and methodology of science. We show the relations of the concepts of “artificial” and “natural,” “strong” and “weak,” “general” and “narrow” intelligence. The article reveals the theoretical difficulties associated with a clear definition of the properties of general intelligence and the ways of its practical implementation. The author draws attention to the importance of studies of consciousness for the development of general artificial intelligence and comes to the conclusion that the priority issues are using the results of the phenomenological analysis of subjective reality, its value-semantic and op-erational structures. The paper discusses these issues in detail, as they are of direct relevance to the construction of new cognitive architectures. The latter make it possible, to go beyond the limits of “narrow” artificial intelligence and create AGI with a high degree of autonomy and independent solutions to a wide range of problems in different environments. It demonstrates the limitation of Turing's operationalist methodology that excludes the use of the results of special studies of consciousness as a subjective reality. Application of such results is associated with the development of post-Turing methodology, which opens up significant opportunities for creating an AGI.

Leibniz defined mathematics as the “science of possible worlds.” But what worlds were assumed as the possible ones by thinkers of different eras? It is this question that is considered in this article. According to Einstein, development of science requires “external justification” (observations and experiments that should be explained or predicted) and internal perfection (following the internal logic of this discipline). Mathematics has no “external justification” and experiment cannot refute mathematical theory. In this sense, mathematics is closer to creativity, to art than to natural science. Its connection with culture turns out to be more complex and mediated. To explain this connection, Daniel Bell, within the framework of his theory of social development, puts forward the “axial principle,” according to which the role of science is considered as the main characteristic of society. From this point of view, we single out traditional, industrial and post-industrial societies. Each of these phases has its own ideals, norms, and types of mathematical creativity. In traditional society, following the Pythagorean tradition, mathematics is focused on finding harmony in nature, on identifying unity on the basis of universal relationships determined by the numerical characteristics of the objects under study. As the industrial era approaches, the constructivist, or “project-based,” approach becomes increasingly important. And unity emerges at a higher “meta-level.” The forerunner of this direction is Descartes, who raised the question of finding a single, universal method for solving all mathematical problems. The work traces the change in the formulation of a number of “invariant,” “eternal” mathematical problems as well as the evolution of the concept of “complexity” in the historical retrospective. Close attention is paid to the post-industrial phase of civilizational development and to “computer mathematics,” which has become the basis for formation of virtual reality, which in many respects changes the very direction of progress. As a result of this, the “extraverted orientation” of humanity, the course toward new horizons was replaced by the “introvert” one, which prioritizes tasks associated with comfort, convenience, and consumption. The “change of milestones” that has taken place is traced in a comparison of “big projects” related to mathematics that were put forward in the 1960s, and those that are considered as priorities at present. In fact, we are faced with a “crisis of expectations.” We see a way out of this crisis in a revival of the “Pythagorean tradition,” at a new level. But if in a traditional society the goal of the development of mathematics in alliance with other sciences and arts was to reveal harmony in the natural world, then at the post-industrial phase, the priorities are different. They are associated with computer modeling, understanding, and identifying the foundations of harmony in the human world.

Neurophilosophy is understood as different areas of philosophy, for example, the philosophy of neuroscience, the philosophy of artificial intelligence, or eliminative materialism. This excessive interpretation of the term is due to the fact that the understanding of the subject area of this discipline is still incomplete. For example, one of the earliest definitions of neurophilosophy given by P.S. Churchland stated reduction of psychology to neurosciences. In modern views, the idea of neurophilosophy as an attempt to justify eliminative materialism is outdated and does not correspond to reality. The article analyzes the terms “philosophy of neuroscience,” “neurophilosophy,” and “philosophy of artificial intelligence” and also offers a variant of their differentiation. The authors focus on the common and different features, using the example of G.M. Edelman's theory of consciousness and the concept of connectionism for weak artificial intelligence. It is concluded that integral use of the term “neurophilosophy” should be abandoned. As a result, the term “neurophilosophy” should be understood as a direction in philosophy of the early 21st century, applying neuroscientific concepts to solve traditional philosophical problems, while the philosophy of specific neurosciences can be considered primarily as a field in the philosophy of science that formulates and solves problems of specific neurosciences as well as of the entire neuroscientific direction. The philosophy of artificial intelligence is an area in philosophy that answers the question of what non-biological intelligence is and what makes it possible; in other words, it is a philosophical and methodological basis for the study of non-biological intelligence. In the formation of neurosciences and their scientific and philosophical basis, we are still at the first methodological stage of the analysis and differentiation of hypotheses. After some time, there will emerge a philosophy of neuroscience, as the basis of all existing neuroscientific theories, and then this term will acquire greater significance.

The article discusses unresolved problems and limitations that arise with application of artificial intelligence (AI). These problems are largely related to the fact that ideas about AI are often formed without taking into account the control paradigms. The most common ones are paradigms that consider artificial intelligence not as means included in control activities or control paradigms, but as independent objects of research in the paradigms corresponding to the specifics of such objects. Such paradigms contribute to the development of certain areas of AI, but they also complicate their application in control processes and ignore many potential areas of AI that are relevant to the development of control problems. The organization of control processes is based on their specific paradigms (subjective, cybernetic, etc.) that set such specific requirements to AI implementations as well as to tasks in which it is advisable to use AI. Such control paradigms form tasks for AI, which contributes to successful practical application and development of AI as well as to mechanisms for controlling and neutralizing negative consequences. The author proposes a mechanism for interaction of subjects (persons) and active forms of AI (considered as pseudo-subjects). Taking into account the increasing role of reflexive activity in the processes of social control, the article considers the place and role of AI in ensuring reflexive activity in the subject paradigms of control. Analysis of trends in the development of controlling from the standpoint of the development of scientific rationality (classical, non-classical, and post-non-classical) allows us to conclude that each subject paradigm of control (“subject - object,” “subject - subject,” and “subject - meta-subject”) has its own specifics, which should be considered when developing active forms of AI.

Over the past decades, due to the course towards digitalization of all areas of life, interest in modeling and creating intelligent systems has increased significantly. However, there are now a stagnation in the industry, a lack of attention to analog and bionic approaches as alternatives to digital, numerous speculations on “neuro” issues for commercial and other purposes, and an increase in social and environmental risks. The article provides an overview of the development of artificial intelligence (AI) conceptions toward increasing the human likeness of machines: from the key ideas of A. Turing and J. von Neumann, who initiated the digitalization of society, to discussions about the definition of AI and the emergence of conceptions of strong and weak AI. Special attention is paid to the approach of A. Sloman, to ideas about the architecture and design of complex artificial systems are considered, which make it possible to “emotionally” expand the idea of weak/strong AI. In the article's section on the necessity and possibility of incorporating emotions into the architecture of AI, the authors reveal the goals and methodological limitations for creating an emotional artificial agent. In addition, the article briefly presents the main principles of the authors' architectural approach to the creation of emotional intellectual systems on the example of the cognitive-affective model of architecture, which allow modeling the impact of emotions on the cognitive processes involved in decision-making processes. The described architectural approach to modeling intelligent systems can be used as a conceptual basis for discussing and formulating a strategy for the development of neurocomputing, philosophy of artificial intelligence, and experimental philosophy, for developing innovative research programs, formulating and solving theoretical and methodological problems.

Today artistic perception of the world and scientific and technical understanding of reality remain the main forms of creative self-realization. For many centuries, starting in Antiquity, art and science went hand in hand in cultural history. However, during the heyday of technogenic civilization, there occurred a split, and since the second half of the 20^th century these two parts of a previously common culture became poorly compatible. According to the authors, the era of digitalization is going to completely replace the algorithmic and instructive professions in technology and service personnel with robots and artificial intelligence, and a person will have to develop the spheres of right-brain practices, which undoubtedly include scientific and technical creativity and art. Like all previous information revolutions, the modern digital revolution is creating new network structures of fast communication and ultra-long-distance, global order. Distance does not matter any more. In this situation, the cognitive maps of a person change radically, new types of self-organization and socialization appear, and there occurs a deformation of value spaces and worldview guidelines. Life in the uncertainty of this new world can only become effective in interaction of man and artificial intelligence, and the only aspect of activity that cannot be replaced by artificial intelligence is creativity and aesthetic experiences as well as culture. Thus, the authors conclude that a new synthesis in culture is needed, the ways of which are discussed in detail in the article. Among them are promotion of new concepts of Science-Art, application of post-Turing methodology, and use of quantum-synergetic anthropology, which develops new ideas about theatrical and engineering creativity, – and these will remain unattainable to artificial intelligence in foreseeable future.

The paper considers the main stages of the development of artificial intelligence (AI). The aim of the paper is to reveal the influence of computerization on social life at present and in the future. The historical analysis method was used to identify the main stages in the development of neural networks and to compare some predictions and real achievements in the field of AI modeling. The article identifies three main periods, widely known as the first and second spring of the era of artificial intelligence and the “winter of AI” – a period of stagnation. The earliest promising developments ran into technical difficulties. At the first stage, the developers tried to copy nature, that is, the neurons of the natural brain, and the tasks turned out to be too complicated. In addition, the first algorithms for training neural networks encountered fundamental difficulties, such as exponential decay of the error vector when propagating into the inner layers of the neural network. It took years to find new algorithmic solutions, to accumulate data for training neural networks, for the next technological breakthrough, known as the “second spring of AI.” The result of the analysis showed that predictions of AI capabilities were often more modest than actual achievements. In the present paper, we make new predictions, according to which computer technology is going play not only an instrumental role but will also have a significant impact on the ideological values of society. This will happen not only due to increasing automation but also due to substitution of creative professions by machines, which will increase the segregation of society and the trend towards individualism. The problem of realism and the search for criteria of truth will also gain importance as self-programmable neural networks evolve.

The summary discusses the history, structure, and areas of activity of the Saint Petersburg branch of the Scientific Council on the Methodology of Artificial Intelligence and Cognitive Research under the Presidium of the Russian Academy of Sciences. The members of the Saint Petersburg branch are focused on practice, which is due to the predominance of researchers in the engineering and natural sciences. According to the author, the Council stimulates work related to interdisciplinary synthesis and convergence of the humanities and natural sciences in solving the problem of artificial intelligence.

The summary presents the main results of the work of the Samara branch of the RAS Scientific Council for the Methodology of Artificial Intelligence and Cognitive Research, created in 2007 on the basis of S.P. Korolev Samara National Research University (Samara University). The Samara branch of the Council and the Samara University held international conferences on information technology, information society, science fiction, established Artificial Intelligence Center as well as completed interdisciplinary technical and humanitarian research projects in the field of socio-humanitarian cybernetics, digital models of creative processes, computational aesthetics.