Can Artificial Intelligence be Conscious?

EasyChair Preprint 13632

Abstract

All magic and mystery disappear as soon as an obscure mysterious concept gets a rigorous formal definition. In order to provide an opportunity to talk about the applicability of philosophical / cognitive concepts to the subject area of AI, it is necessary to "ground" these concepts by formulating rigorous formal definitions for them. The fundamental importance of such formal definitions is quite obvious, since any concepts applied to the field of Information Technology must be "codable", i.e. potentially implementable in program code. Thus, the "codable" formal definitions of cognitive terms are the necessary basis on which alone it is possible to build the architecture of AI technology that has the ability to embody these concepts in a real software. The question of the adequacy of such definitions of "reality" and their compliance with existing generally accepted philosophical theories is also very important and quite discussable, but this does not affect the priority and fundamental nature of the requirement for the formulation of "codable" formal definitions. The formulation of "codable" definitions for the concept of "consciousness" and related cognitive concepts and, based on them, statements about their applicability to the subject area of AI is the topic of this publication. Covering questions: Can AI have a Personality / Motivations / Free Will?

Keyphrases: AGI, AI, Consciousness, HLAI, LLM, motivation, perception, subjectness, understanding

@booklet{EasyChair:13632,
  author    = {Victor Senkevich},
  title     = {Can Artificial Intelligence be Conscious?},
  howpublished = {EasyChair Preprint 13632},
  year      = {EasyChair, 2024}}