Transferring human axiology to artificial intelligence

This article addresses a key problem in contemporary technoscience: how to effectively implement human values in artificial intelligence systems. The author analyzes the fundamental ontological gap between the syntax of code and the semantics of human axiology. The text provides a detailed discussion of the concept of coherent extrapolated will (CEV) and Nick Bostrom's orthogonality thesis, pointing out the risks of a 'treacherous turn' and the paperclip scenario. The article also addresses the problem of direct norm specification and the ontological crises arising from the evolution of autonomous systems. This is a profound analysis of machine ethics that illuminates the difficulties of creating intelligence that is safe and consistent with human intentions, emphasizing the role of the sovereign sense in the design of future technologies.