Artificial Intelligence Changes Cognition Systems

Purpose of the study. Humanity is at a turning point in its development, which is reflected in the change in the regimes of global processes, the growth of global hypercompetition for the future. Digital transformation is also entering a new phase. The processes of automation, digitalization, and the formation of intellectual production are gradually transforming, and technological development is being replaced by socio-technological development. The world changing before our eyes is becoming more complex, knowledge about it goes beyond the set of individual disciplines and continuously develops as innovative digital transformations grow, including the formation of artificial intelligence. The purpose of the article is to formulate humanitarian problems in the implementation of AI technologies in science and to develop possible directions for their solution. Materials and methods. The research was carried out based on a study and a systematic review of the literature, using methods of comparative analysis and typologization, analysis of epistemic risks based on the synthesis of analyzed sources. The use of the conceptual modeling method made it possible to describe the iterative “datamodel-experiment” cycle as the basis of the Fifth paradigm and justify the key role of understanding in this process. Research results: the analysis made it possible to systematize modern ideas about AI scientists, as well as to highlight the fundamental limitations, risks and prospects for their integration into scientific practice. The article, based on the works of foreign researchers, reviewed the key issues of the use of AI in science and the development of AI scientists. The current trends in the development of AI technologies that form the use of artificial intelligence for scientific discoveries are analyzed, the best ways to use AI and the problems associated with them are considered. Key limitations and problems of the “AI-scientist” have been identified. General ideas and practical AI solutions that affect the cognition system are described; fundamental limitations and key problems in the use of AI scientists are reflected. Conclusion. Monitoring new knowledge, identifying, creating and storing key information and data from exponentially growing streams have become impossible without highly developed artificial intelligence. The fourth scientific paradigm actually ended: big data gave way to a cognitive system that organizes an interactive research process an iterative cycle of creating new knowledge. The model created on the basis of data during the experiment provides new data, further analysis of which, supplemented by expert knowledge and elements of interdisciplinary knowledge, allows it to improve and refine the initial hypothesis. Thus, the processes of cognition are continuously accelerated and changed. The study confirms that the world scientific community is at the point of bifurcation associated with the transition from the Fourth paradigm (based on big data and computing) to the Fifth paradigm, in which the central link is the cognitive system that implements the iterative cycle “data model experiment new knowledge”. Generative artificial intelligence (GenAI) and large language models (LLM) act as catalysts for this transition, radically changing the methodology, speed and very nature of scientific search. The authors believe that new approaches and solutions can significantly advance the creation of AI capable of scientific reasoning. However, when working with AI scientists, it is necessary to take into account epistemic risks, as well as the fact that the landscape of AI technologies is changing extremely quickly, which requires constant updating of the obtained typologies and conclusions.