Part of the change in contemporary mourning and memorial culture is that its rituals are no longer only performed in the analogue world, but increasingly also in the digital world. On the one hand, this opens up opportunities for emotional accompaniment, support and connection across physical distances, and on the other hand, new variants of the social presence of the deceased are emerging. The preservation of personal (analogue and digital) traces that provide information about a past life and thus give the dead a permanent appearance is not a new phenomenon in principle. However, with the help of artificial intelligence (AI), people are now being offered the prospect of an unprecedented form of interactive existence after death. Within the population, this not only arouses curiosity and fascination, but also doubt and unease. The associated questions relate to current and future interpersonal coexistence and how to deal with new technologies, about which there is still relatively little knowledge.
Media of presence
The need for a sensually tangible presence beyond the physical end of life has a long cultural history, in the course of which different practices, techniques and media have been used time and again. The spectrum ranges from cave paintings, sculptures and paintings, the spoken and written word to modern image technologies such as (digital) photography and videography. The fact that more and more people are producing more and more digital data in the course of their lives, which can reveal something about them and potentially outlive them, is also playing an increasingly important role in the way in which they mourn and remember each other. Although cemeteries continue to be of great importance to many people as traditional places of mourning and remembrance, a trend towards delocalisation has been emerging for some time. This refers, on the one hand, to the general detachment from previously binding spatial fixations and, on the other, to a general loss of relevance of the dead body and its localisation in the context of mourning and remembrance. This development is further accelerated by the emergence of digital offerings and their advantages.
For around 30 years, a growing interweaving of digital technologies with mourning and remembrance practices has been observed - for example in the form of virtual cemeteries, online memorial pages or specific mourning forums. These now also include various social media platforms where, in addition to countless other life topics, dealing with dying, death and mourning is also dealt with in words and images. Profile pages of deceased users are sometimes transformed into virtual memorials, as other people continue to visit them (or visit them all the more frequently) and add specific content related to mourning and remembrance. There are now numerous videos on YouTube that users have created in memory of deceased friends or family members. In addition, customisable virtual spaces of remembrance are offered in which users can 'immerse' themselves and encounter various traces of the deceased (for example in the form of photos, video and audio recordings or digital replicas of personal items).
Digital afterlife industry services go one step further, utilising artificial intelligence (AI) technologies and promising post-mortem survival as an avatar that can communicate with the living. This involves simulating a person's communication behaviour, personality traits (attitudes, beliefs, taste preferences, hobbies, etc.) and therefore also their physical appearance based on huge amounts of digital data they have left behind (e.g. emails, messenger histories, social media posts, calendar entries, photos, videos, voice messages, etc.). In addition to comparatively simple forms in which the previously stored original material is selectively selected during use but output unchanged, there are also AI systems that generate new output from the underlying data of the deceased and bring it into the interaction with the user. Accordingly, the avatar says things that the person it represents never said, but could say in this way if they were still alive.
Forms of digital survival and their ethical consideration
The AI-based technologies of digital survival have a wide range of applications. One of these is popular culture: while deceased artists such as Michael Jackson or Elvis Presley have already celebrated their post-mortem stage comeback as virtual animations, actors and actresses can be represented or replaced by their digital doubles. Other public figures have also experienced a certain continued existence in the digital world. On 22 November 1963, then US President John F. Kennedy was due to give a speech in Dallas, Texas, had he not fallen victim to a fatal assassination attempt shortly before his arrival. More than half a century later - in 2018 - the speech in question was set to music with Kennedy's synthesised voice. Other applications can be found in the field of historical and political education, where the digital representations of Holocaust survivors answer users' questions in museums, classrooms or on the display of a personal device. The eyewitnesses in question were interviewed extensively in advance on hundreds of questions and filmed by several cameras from different perspectives. The resulting material can be processed using speech recognition software so that the answer sequence matching the question asked is played back in the style of a live conversation.
In addition to the digital survival of publicly known people, mourning and remembering people from the private sphere also offers a lucrative line of business for the digital afterlife industry. For example, a smartphone app sends bereaved people text, voice or video messages from the deceased on special occasions or after they have entered a specific question. In the UK in 2022, shortly after the funeral of an elderly lady, her digital counterpart was displayed on a large screen and answered questions from the assembled mourners. Another example comes from South Korea, where a TV show showed a woman's encounter with a three-dimensional simulation of her deceased seven-year-old daughter in a virtual park.
Communication with their AI simulations helps to continue the relationship with the dead and create the feeling of their continued presence - at least that is what the marketing rhetoric of individual providers promises. One of them even advertises with the slogan "Never have to say goodbye". Whether such a decoupling of death and farewell is a promise or a threat is open to debate. After all, digital survival in this sense harbours quite a few problems that are still largely unresolved and touch on a number of ethical dimensions in addition to questions of personal rights and data security: For example, how can the privacy of the deceased and their surviving relatives be adequately protected? Where exactly does the (and therefore highly sensitive) data required to design the avatar come from and who selects it? From the deceased themselves, from their relatives or, for example, from the companies that provide the service and whose economic interests are not necessarily in the service of successfully coping with loss?
To what extent is the digital 'after' not per se permeated by commercial aspects of profit maximisation - and what risks of manipulability does this pose for interaction with the avatars? The scenario in which the digital version of the deceased grandmother advises her grandchildren to buy a certain product may seem somewhat bold, but such, and therefore more subtle, forms of influence cannot be ruled out in economic terms.
These and other aspects are the focus of a research project on the ethics, law and security of digital survival (2022-2024) conducted at the University of Tübingen (in cooperation with Fraunhofer SIT in Darmstadt) and funded by the BMBF. It focussed on the overarching question of the extent to which the technological change described above is changing society and social relationships, particularly with regard to the finite nature of life. Experts from various fields of activity relating to dying, death and mourning (e.g. from the funeral service, certain religious communities, death and bereavement counselling, etc.), as well as private individuals and developers of specific IT services, were asked for their assessment of the aforementioned variants of digital survival through AI.
The respondents were predominantly sceptical. There are major concerns, particularly with regard to possible effects on the grieving process. While the realisation and recognition of the irretrievable loss is seen as an essential prerequisite for successfully coping with grief, the ongoing interaction with an avatar that gives the impression that the missing person is still alive could prevent precisely this. This could allow users to escape into a kind of illusory reality, which could also be accompanied by a withdrawal from the social environment of the analogue world. Some of the study participants even compare this to the consumption of a drug, which numbs emotional pain in the short term, but brings further, sometimes even more serious problems in the long term. There is also increasing concern that personal memories of the deceased person could be overwritten by alternative content produced by the AI system, which is difficult to understand. The latter is thus seen as having a manipulative potential that could have a negative impact on the well-being of the mourners and their relationship with the deceased. For example, the avatar could spread untruths about their analogue role model or make statements that hurt the bereaved. Their emotional dependence would in turn make it more difficult for them to turn away from the service or cancel it, as such a step could be interpreted as a "second death" of the deceased person.
Future application scenarios
Due to the low prevalence and use of such applications in the private sector to date, there is currently still a lack of reliable empirical findings on their actual effects on the grieving process of relatives. Nevertheless, it is possible to anticipate some factors that could play a role in this regard: for example, the age of the deceased, the circumstances surrounding their death, the quality of the relationship perceived by the bereaved and the timing and duration of the use of the AI simulation. The specific expectations of the users are also likely to be decisive: What do they specifically expect from this or that service? Are they really interested in recreating a person in all their personality complexity? Are they looking for a permanent exchange with an avatar that is available at all times in order to continue the shared history of communication with the deceased person more or less without interruption? Or is it rather temporary reminiscences that motivate someone to use a particular digital service? There is no need to emphasise that grief is not a static state, but is subject to a certain dynamic and that needs in this regard - also with regard to appropriate forms of remembrance and interaction - can change over time. Irrespective of this, the question should also be asked as to whether coping with grief and loss is always central in the context of the digital survival of private individuals. Finally, it would also be possible to imagine applications that are less laden with meaning and are easier to integrate into everyday life due to their more playful nature. Here, for example, certain voice assistance systems could be considered, in which the familiar voice of a deceased person could be heard instead of that of a stranger.
The forms of digital fortexistence through AI discussed in this article call for some fundamental considerations. These concern, among other things, the question of the definition of liveliness beyond biological processes and corresponding shifts in boundaries. This seems particularly remarkable in a society that is described as secularised, but in which at the same time there is a persistent longing for transcendence. It is also worth discussing what constitutes human personality and interaction at its core and to what extent all of this can be adequately depicted in the quantities of data left behind.
Certainly, development is still in its infancy and many services (especially those aimed at private individuals and specialising in the creation of new content using generative AI) are currently in a test version and have only been used by a few people so far. However, neither the current technical limitations nor the reservations expressed in the study should ultimately change the fact that the services in question are likely to become even more widespread. In addition to technological change, demographic change is also a key driving force behind this. The majority of those who die each year still spent the majority of their lives in a time before the internet became an omnipresent everyday medium. But what will happen in a future society that will consist almost entirely of 'digital natives'? A person's digital legacy could then be based on seven, eight or even nine decades of online existence - and provide a promising basis for a convincing AI simulation after the physical end of life.
Seen in this light, the fact that a field of experimentation for individual technology pioneers can become a versatile tool for a broad user community is perhaps also simply a question of cultural familiarisation with the fact that 1) co-presence is increasingly mediated digitally, 2) interaction with artificial virtual persons has real emotional consequences and 3) life courses receive digital manifestations that are effective beyond death. In view of this development potential, it seems all the more important to start thinking about its social, cultural and ethical implications at an early stage rather than in the future. 
