ETHICAL CODIFICATION FOR STRONG ARTIFICIAL INTELLIGENCE AS A DILEMMA OF “IMPOSSIBLE CHOICE”

The purpose of the article is to critically analyse the problem of creating an ethical codification for strong artificial intelligence algorithms. The scientific novelty lies in the consideration of two images of AI based on the postmodernist interpretation of language as a discourse of power and the concept of the ‘impossible’, forced choice, by making which a person risks losing the very ability to choose (A. Zupančič). The first image is as an alienated Other (Doppelganger), born within the natural mechanisms of evolution of the theme of monologic progress (rationality), with its inexorable logic of (self-)destruction, which allows us to talk about ethical codification. And in this case, codification will be represented by the newspeak with its paradigmatic intension of suppression/self-limitation/self-negation, will act as a self-reference of human destructiveness, realised as autodestruction in its purest form. The authors proceed from the assumption of the reflection of this schematic in the human feeling of threat - fear of AI and, at the same time, in the annoying insistence of institutional codification as an illusory way to contain alienated destructive forces, carried out by the tools of the ‘terror ethics’, withdrawing and accumulating the substrate of subjectivation. Subjectivation, represented by the ‘burden’ of the now alienated, responsible choice in the sphere of a new AI ontology, where the unconditional is trapped in a ‘new’ conditionality (Jaspers). The second image is a peculiar apophatic project of artificial intelligence, in which it (AI) exists as a productive/producing constantly changing entities mirror recursion (Solaris), eliminating the subject's substanceality and enabling superpersonalisation - purification of choice from determinations by conventions/particularities. The result of the study is the identification of a common dilemma of the necessity/non-necessity of ethical codification for strong AI (the paradox of impossible choice), on the basis of which the authors distinguish a number of specific ethical dilemmas in the sphere of language and control: 1) creation of a ‘personal’, non-transparent for humans language of algorithms as their freedom and autonomy / universal language for all actors of communication; 2) exclusivity of the human mind / recognition of its limitations; 3) a single ultimate moral censor establishing a strict set of moral regulators / pluralistic, situational morality in the relational space.