Review of Lemma, A. (2024). “Mourning, melancholia and machines: An applied psychoanalytic investigation of mourning in the age of griefbots”. The International Journal of Psychoanalysis, 105(4), 542–563. https://doi.org/10.1080/00207578.2024.2342917

In psychoanalytic theory the experience of loss is central to the development of psychic structures.The centrality of mourning to psychoanalysis motivates Alessandra Lemma's paper in which by analysing an episode of a science fiction TV show, she critically evaluates the use of AI chatbots in managing grief after the loss of a loved person.In an episode of the Charlie Brooker's dystopian series Black Mirror entitled BRB (be right back) we meet Martha and Ash, a young couple very much in love.They cherish each other's quirks, playfully teasing each other.Then suddenly Ash dies in a road accident and Martha is inconsolable.A well-intentioned friend subscribes Martha to an AI chatbot designed to simulate conversations with the deceased.It uses natural language processing and machine learning algorithms to mimic the personality, expression, and speech patterns of someone who has died.This is achieved by mapping the individual idiom of the lost person based on online behavioural samples including phone and text conversations, emails, videos, and social media posts.Initially Martha recoils from the idea but because the chatbot is able to analyse and adapt to the user's behaviour it reassures Martha.She enters the illusion and becomes dependent on continuous contact with "Ashbot", withdrawing from the world around her. Ultimately Martha cannot bear the fact that while Ashbot can replicate Ash it cannot replace him, crying out "You're not enough of him.You're just a few ripples of him."The premise of the Turing test was whether a computer talking with you through text on a screen could convince you that you were talking to another person.With the advent of generative AI, this milestone has been passed.All the leading AI models could handily impersonate a human Ph.D. candidate in an oral thesis defence.To use an AI in this way would be ethically wrong, but what are the implications for other ways in which AI may be used to simulate human interactions?