Reshaping Grief with AI: What would be the Consequences of Introducing Griefbots as Supports for Bereaved People in the Mourning Process?

Abstract

Over the past decade, the internet has seen a proliferation of AI companies promising to enable bereaved people to create griefbots of their deceased loved ones. Griefbots are generative AI conversational algorithms fine-tuned on digital remains of the deceased to imitate them. The philosophical debate about griefbots follows two main lines of thought. One sees griefbots as potential transitional objects that the grieving person can integrate into their grief process, keeping a continuing bond with the deceased. The other has raised understandable ethical concerns about these technologies. The focus of my analysis will be on the consequences of the use of griefbots on the autonomy, health and dignity of the bereaved. In section 2, I will provide a definition and comparison of the Freudian relinquishment theory and the continuing bonds perspective. In section 3, I will summarise both lines of thought about griefbots. In section 4, I will express my concerns about the use of these technologies in the grief process. First, I will explain whether griefbots can be labeled as transitional objects compared to other affective artifacts, based on Winnicott's definition of transitional objects. Secondly, I will analyse the possible risk of delusion and self-deception for mourning users when interacting with griefbots. Then, I will analyse the potential risk for the bereaved of isolation and detachment from reality in the interaction with griefbots. In the conclusions, I'll list some of the solutions that have been proposed to protect users' health and agency.