Understanding grieving for a chatbot using two concepts from Wittgenstein

In February 2023, changes to the AI chatbot Replika elicited a grieving response from a number of its users. This paper examines how some users of Replika expressed their sense of loss: in particular, those who write as if the chatbot was a subject in its own right. It is easy even for those who feel some sympathy for these users to dismiss the way in which they express themselves as inappropriate, on the grounds that the entity for which they are grieving is not a person. This paper argues against this dismissal, making use of two concepts from Part II of Wittgenstein’s Philosophical Investigations. It argues that what is distinctive about interaction with a chatbot is their ability to stimulate in users something not unlike (one dimension of) what Wittgenstein describes as “an attitude towards a soul”: that is, a distinctively responsive bearing towards the chatbot. The paper argues that a user of an AI chatbot might, for this reason, feel that the only language suitable for expressing their experience is the language we would use in grieving a person, even though they know the chatbot not to be a person. That is what Wittgenstein calls a secondary use of that language. Concerning the character of the experience reported, this is revealing: it stands in a parasitic relationship to cases of grief for a real person.