The Moral Turing Test within the Frameworks for Normalizing Attitudes towards AI in Socially Significant Future Technologies

Abstract

The proliferation of new technologies raises the problem of applying and adapting the Turing Test to evaluate the moral decisions made by artificial intelligence (AI) systems in the context of bioethics. The relevance of this problem for the philosophy of culture lies in the need to analyze the prospects for the harmonious coexistence of humans and artificial systems, considering dominant cultural normative systems, one of which is morality. The aim of this research is to refine approaches to solving ethical problems associated with AI against the backdrop of its integration into the latest social technologies. The research objectives were as follows: 1) to identify and describe the problems associated with the spread of AI in the social sphere; 2) to clarify the specifics of the ethical questions arising from the implementation of AI in this area; 3) to systematize knowledge about existing deontological frameworks that aim to address the problem of the social normalization of AI use. The research materials include information on the latest developments in social engineering, namely technologies that apply AI to solve social tasks (in medicine and elderly care), as well as scholarly literature devoted to the use of AI in the social engineering of the future. The study is based on a culture-oriented approach. The methods used involve case analysis and SWOT analysis. Based on the analysis of scholarly literature, various modifications of the Moral Turing Test are presented: the Comparative Moral Turing Test (cMTT), the Ethical Competence Test, the Machine Ethics Safety Test, and the Turing Triage Test. As a result of the research, it is shown that the Moral Turing Test is a functional tool for demonstrating the ethical safety of artificial systems but cannot serve as proof of their possessing moral agency in the human sense, which is particularly relevant for the sensitive sphere of bioethics. The study concludes that: First, within the framework of developing the aforementioned modifications, the methodological difficulties and fundamental limitations of these approaches are described. These include the problem of imitation, the "absence of understanding" in AI, the risk of software errors, and the fundamental differences between thinking and the capacity to be a moral agent. Second, the practical significance of developing criteria for the ethical verification of AI is demonstrated, and the specific bioethical problems arising from its use are clarified (the problem of responsibility, patient autonomy, stigmatization, and equality of access). Third, philosophical approaches to the question of the possibility of creating "genuinely" moral AI are systematized; objections against this thesis are highlighted, based on the arguments of biological naturalism (J. Searle), phenomenology (H. Dreyfus), as well as the concept of the erosion of human moral skills.

Ähnliche Arbeiten