Understanding Sophia? On human interaction with AI systems and robots

CFS Lecture by Thomas Fuchs, Karl Jaspers Professor for Philosophy and Psychiatry at the Department of General Psychiatry, Universität Heidelberg.

Abstract

Against the backdrop of increasing similarity between robotic performance and human communication, the paper explores the question of whether there can be shared meaning constitution and empathy in human-AI and human-robot interaction. It first describes the human tendency to anthropomorphize artificial agents and then analyses the concept of understanding, which presupposes (a) the subjectivity of the counterpart, (b) shared feelings, and (c) shared intentions or a "we-intentionality". Furthermore, the assumption of a possible subjectivity realized by future AI systems is rejected on the basis of an enactive account of consciousness. Finally, the risks of "digital animism" are pointed out, illustrated by the example of virtual or chatbot psychotherapy. In summary, the paper's main contribution to research lies in a close examination of the intentional terms often applied to AI and an exploration of the category errors and anthropomorphisms associated with them.