Interactions between a human and a Chat Bot: The mythopoietic narrative of unlimited possibilities

In the debate on the Large Language Models (LLM) there are three arguments that fuel the mythopoetic narrative concerning the emancipation of the artificial agent vis-à-vis the human being: 1. The conscious self-reflexive process; 2. The ability to deeply understand the 'semantics' of natural language; 3. The creative-transformative ability. These arguments about the future self-emancipating possibilities of A.I. are being advanced by transhumanism and the current known as “Strong A.I.”. However, such narratives have been disproved by a different concrete reality in which the interaction between human and chat bot remains asymmetrical: on the one hand we have a subject endowed with meaning and intentionality that is capable of changing and inventing content in an unpredictable way ; on the other hand we have an expert system that assembles content and combines expressive possibilities without fully understanding what it is doing. This article will falsify the mythopoetic construction by examining the prediction paradox, the Inverse Turing test, and the algorithmic hallucinations