Souls and Selves: Querying an AI Self with a View to Human Selves and Consciousness

Abstract

The question of self-aware artificial intelligence may turn on the question of the human self. To explore some of the possibilities in play we start from an assumption that the self is often pre-analytically and by default conceptually viewed along lines that have likely been based on or from the kind of Abrahamic faith notion as expressed by a “true essence” (although not necessarily a static one), such as is given in the often vaguely used “soul”. Yet, we contend that the self is separately definable, and in relatively narrow terms; if so, of what could the self be composed? We begin with a brief review of the descriptions of the soul as expressed by some sample scriptural references taken from these religious lineages, and then transition to attempt a self-concept in psychological and cognitive terms that necessarily differentiates and delimits it from the ambiguous word “soul”. From these efforts too will emerge the type of elements that are needed for a self to be present, allowing us to think of the self in an artificial intelligence (AI) context. If AI might have a self, could it be substantively close to a human’s? Would an “en-selved” AI be achievable? I will argue that there are reasons to think so, but that everything hinges on how we understand consciousness, and hence ruminating on that area—and the possibility or lack thereof in extension to non-organic devices—will comprise our summative consideration of the pertinent theoretical aspects. Finally, the practical will need to be briefly addressed, and for this, some of the questions that would have to be asked regarding what it might mean ethically to relate to AI if an “artificial self” could indeed arise will be raised but not answered. To think fairly about artificial intelligence without anthropomorphizing it we need to better understand our own selves and our own minds. This paper will attempt to analyze the self within these bounds.