Five reasons why a conversational artificial intelligence cannot be treated as a moral agent in psychotherapy

Abstract

Sedlakova and Trachsel present an analysis regarding the evaluation of a new therapeutic technology, namely conversational artificial intelligence (CAI) in psychotherapy. They suggest that CAI cannot be treated as an equal partner in the therapeutic conversation, because it is not a moral agent. I agree that CAI is not a moral agent. However, I believe that CAI lacks at least five basic attributes or abilities (phenomenal consciousness, intentionality, ethical reflection, prudence, conscience) that would allow it to be defined as a moral agent. It seems that the ethical assessment of the possibilities, limitations, benefits and risks associated with the use of CAI in psychotherapy requires a determination of what CAI is in its moral nature. In this paper, I attempt to show that CAI is devoid of essential moral elements and hence cannot be treated as a moral agent.