Проблема ориентации искусственного интеллекта на человеческие ценности (AI value alignment) и социология морали

Социологические исследования. 2023. № 9. С. 16–28.

Deviatko I. F.

The article briefly examines the popular ideas about the goals and possibilities of human control over artificial intelligence that has been developed at the earlier stages of the scientific and technological revolution and substantiates the thesis about their incompleteness in terms of not taking into account new asymmetries of control and technological realities that arose as a result of the "digital revolution". An analysis of the reasons why the sociology and social psychology of morality are acquiring a decisive role as well as a new large-scale research field in the development of ethically oriented AI systems is presented, which reconfirms the importance of a theoretically based empirical study of the normative dimension of social life. An additional sociological substantiation and a narrower interpretation of the AI value alignment principles put forward by some authors as a solution to the problems of the ethical orientation of AI systems is proposed.

Keywords: искусственный интеллект, справедливость, justice, социология морали, sociology of morality, artificial intelligence, множественность нормативных систем, AI value alignment, plurality of normative systems, совмещение ИИ с человеческими ценностями