AI Companionship

Abstract Loneliness is one of the biggest global problems of our era and has severe consequences both for individual health and society at large. AI companions, such as social robots or chatbots, are one technological means increasingly relied on to fend off loneliness. It is argued in this paper that while AI companions could help with easing loneliness, the increased reliance on them to address loneliness, especially in everyday contexts, might undermine the problem it tries to address. AI companions do not have an experiencing living body, so they cannot understand or empathize with human beings; additionally, the human–AI relationship lacks mutual relationality and is characterized by simulated relationality. So, even if AI companions, by behaving empathically and by their simulated relationality, can ameliorate the negative feeling of being alone, they could not substantially address the reasons for loneliness. Despite this, AI companions have many features that exploit human psychology and that appear all-positive for the user, which makes them highly appealing. However, because of these all-positive conditions, and because they are social actors with the power to contribute in shaping social practices, AI companions could transform, for the worse, how humans understand and what they expect out of human–human relationships. Thus AI companions might contribute to more human isolation, which would increase the chances of loneliness, or they could bring about a new type of loneliness, where one is without the negative feeling that accompanies loneliness but still lacks the desired quantity and/or quality of human relationships.