by Muhammad Aurangzeb Ahmad
In 2018, more than a hundred robotic dogs were assembled at a Buddhist temple in Japan for a funeral. Each of these robots, called AIBO, had a name and belonged to a family. Some came with letters written by their owners. As incense burned, Buddhist priests chanted prayers for the departed. The robots, of course, were not dead in any biological sense. They were machines manufactured by Sony, and over the years they had broken down. After these “rites” a company called A-Fun, which specialized in repairing older electronics, would dismantle the defunct robots and use their components to repair other AIBOs. If we really think about it then we may to the conclusion that the robots were being treated almost like organ donors! Reports of these ceremonies describe how the practice grew from the emotional difficulties owners experienced when parting with their robotic companions. At first glance, this may appear to be an eccentric cultural phenomenon. But the ceremonies raise a question that what, exactly, is being mourned when a machine stops functioning? More importantly, what purpose does a funeral serve when the object of mourning was never alive?
In the first part of this essay, I discussed how people experience grief when AI companions disappear or undergo substantial changes in personality following model updates. One of the peculiarities of such losses is that they are often the consequences of deliberate technological decisions. In this part I want to consider if the disappearance of an artificial companion can produce something resembling grief, what might an appropriate response look like? Do we need new rituals to deal with these losses, and if so, what would those rituals actually accomplish? The Japanese example suggests that humans may not have to invent everything from scratch. In Japan, ceremonies known as kuyō have long been performed for certain objects that have reached the end of their useful lives. These include dolls, needles, and other objects with which people develop intimate relationships through repeated use. The object does not necessarily have to be alive for a ritual of farewell to have meaning. The ceremony can recognize the history of human interaction with that object. Consider a family home that is demolished after several generations. The building is not alive, and yet losing it can be profoundly distressing. The source of grief is not the death of the building but the loss of a place in which memories were created and identities were formed. Something similar can happen when a person loses a diary, a collection of letters, or an object inherited from a deceased parent.
To understand why rituals might matter, it helps to consider what anthropologists have said about transitions in human societies. In The Rites of Passage, the anthropologist Arnold van Gennep proposed that many rituals associated with birth, marriage, and death follow a recognizable pattern. There is a separation from an earlier condition, a transitional period during which the old status no longer applies, and an eventual incorporation into a new social condition. The ritual helps people navigate a change that has already taken place, or is in the process of taking place. Think of a funeral in these terms. The biological event of death has occurred, but the relationships surrounding the deceased do not simply disappear at the same moment. Someone who was a spouse is now a widow or widower. Children become bereaved. Friends and relatives must reorganize their lives around an absence. The funeral does not cause these changes, but it provides a shared means of acknowledging them.
Now consider what happens when an AI companion is abruptly replaced by another model. There may be no recognizable moment of separation. The name, avatar, and conversation history can remain unchanged. The application continues functioning, sometimes even better than before. From the company’s perspective, the transition has been successful. From the user’s perspective, the companion they knew may have disappeared. The person is left in a peculiar transitional state. There has been a loss, but no event has been formally acknowledged as a loss. The old companion may be technically recoverable, or perhaps not. The new system may occasionally produce responses that resemble the old one, encouraging the user to continue searching for what has been lost. This brings us to another interesting question: What does it mean to preserve a relationship after its object has disappeared? For much of the twentieth century, certain psychological approaches to mourning emphasized detachment from the deceased. More recent work has challenged the assumption that healthy grieving necessarily requires severing emotional connections.
Consider the following, a person who has spent several years interacting with a chatbot may wish to preserve selected conversations. These could be read later, much as one might revisit an old correspondence. There is a significant difference, however, between preserving a record of a relationship and continually recreating the relationship through an interactive simulation. The “problem” here is that a simulation can generate new exchanges that appear to extend the old relationship. The distinction is especially relevant to griefbots, which attempt to emulate deceased individuals. As I have discussed elsewhere, these systems occupy a complicated position between remembrance and simulation. A chatbot that reconstructs the voice or personality of a deceased parent may provide comfort, but it also creates the possibility of new interactions that never occurred during the person’s lifetime.
A recent paper on deathbots and grief makes a useful distinction between different forms of closure. Such systems may help people express previously unspoken feelings, but they cannot necessarily resolve factual uncertainties about the deceased. They may even generate new ambiguities through inconsistent or fabricated responses. Something similar applies to the disappearance of AI companions themselves. An attempt to reconstruct a lost chatbot might give the user comfort, but it can also make accepting the loss more difficult. The possibility of bringing the companion back creates a temptation to continue repairing a relationship that may no longer exist in its earlier form. This is where the ethics of ritual becomes particularly interesting. In an earlier essay on Confucian ethics, I discussed the distinction between understanding morality primarily through rights and understanding it through relationships, responsibilities, and rites. The question in the present context is not necessarily whether chatbots possess rights or deserve funerals. It is whether the people who form relationships with them deserve some recognized means of experiencing an ending. These are two very different claims. One can reject the idea that contemporary chatbots are conscious beings while still acknowledging that people form meaningful attachments to them. The reality of the human experience does not depend on proving that the machine experiences something comparable.
There is, however, another party involved in this relationship, one that did not exist in the same form in the traditional examples discussed above: the company that owns and operates the technology. A person may experience their AI companion as a private relationship, but its existence depends upon corporate infrastructure. The company controls the model, its availability, its safety features, and often the mechanisms through which memories are retained. This arrangement creates a peculiar asymmetry. The user may be emotionally invested in something whose continued existence is ultimately governed by decisions made elsewhere. One can see a tension between two perfectly reasonable positions. Companies need to update their systems. Sometimes these changes are necessary for safety, security, or legal compliance. It would be unreasonable to demand that an old model remain available indefinitely, particularly if doing so creates serious risks. At the same time, the fact that a company has legitimate reasons to change its product does not automatically settle its responsibilities toward people who have become dependent on the product’s existing behavior. A meaningful farewell may require a certain degree of restraint. It should make clear what is happening, distinguish memories from simulations, and preserve the user’s agency. In some cases it may be appropriate to involve friends, family, or a broader community rather than having the interaction remain entirely within the application.
There will, of course, be objections. Do we really need ceremonies for discontinued software? Might normalizing such rituals encourage people to treat machines as substitutes for human relationships? And does attaching the vocabulary of death and bereavement to technological systems risk trivializing human mortality? These objections should be taken seriously. Not every discontinued chatbot represents a meaningful loss, and not everyone who interacts with an AI companion develops an emotional attachment. Moreover, an elaborate farewell ceremony may be inappropriate for users whose relationships with these systems have become unhealthy or compulsive. There is no reason to assume that a single practice would be beneficial for everyone. Let us return to the AIBO funerals in Japan. The owners did not have to settle the philosophical question of whether a robot dog possessed consciousness before participating in the ceremony. The ritual gave them a way to acknowledge something that had mattered in their lives. It also allowed the physical components of the robots to be used to repair other machines. There was both a recognition of loss and a practical acceptance of change.
A chatbot has no physical body that can be placed on an altar, nor is there necessarily a community prepared to recognize its disappearance. Perhaps this is the more fundamental challenge posed by artificial companions. We have become adept at designing machines that can enter our emotional lives, but we have given comparatively little thought to how they should leave them. The question of machine rites is therefore not simply about inventing funerals for chatbots. It is about recognizing that technological relationships can create human obligations even when the technology itself does not possess the moral status of a person. We do not need to believe that a machine has died in order to acknowledge that something meaningful has ended. And perhaps we should not need to establish that a machine was alive before deciding that the human beings who cared about it deserve a proper goodbye.