‘The digital afterlife industry’, ‘grief tech’, ‘death bots’. The phenomenon has many names. It is a growing global market where companies promise that, for a fee (and sometimes even for free), they can feed an AI with, for example, text messages, emails, voice recordings, videos, or diaries of deceased persons and, based on these, send back interactive avatars of the deceased. The user interface is, for example, an app through which the bereaved can talk or write to a simulated version of their loved one.
Because generative AI can be quite convincing, it can leave some people under the illusion that the deceased is still here. There is also a risk that grief will be prolonged through this simulated, artificial interaction, where the bereaved do not develop alongside their loved one but are solely based on old data that fuels this probability machine, which is also shaped by the bereaved’s prompts.
The rapid development of generative AI has accelerated growth in the market. To my knowledge, there are no estimates of the size of the market in Denmark, but it’s growing globally.
This acceleration raises deeper ethical, psychological, and philosophical questions that are easy to dismiss with a simple: “If the deceased has given their consent, what’s the problem—people should be able to do what they want.” Or: “If it works, what’s the problem?” But first of all, we don’t know if it works. And what is it supposed to work on anyway? The pain? The grief? The loss? The emptiness? The loneliness? And who is it that can ‘do as they please’? Is it the deceased or selected relatives – and if so, whose wishes carry the most weight in this context?
Things are more complicated than that – and I would like to shed some light on this.
Commercialization of Death
In a state of deep grief and pressure, people may be willing to do anything to hold on to their loved ones – one way or another. The industry is already experimenting with different business models, and it is easy to imagine a version that is free for the first period of time – which can initiate a form of dependency – and then upgrade products to a paid version. First fix is free…
Like other companies, commercial grief tech companies are also driven by profit motives and are thus incentivized to create business models that exploit people in mourning. In the grieving phase, all people are the perfect victims.
Let’s imagine that the deceased loved a particular country, a particular dish, a particular wine, or a particular beauty product. It would be an obviously lucrative idea to use the avatar of the deceased to advertise this—or alternatively, to ‘just’ place advertisements for the products in the app. This would probably be an effective form of manipulation.
We know from leaked documents from tech giant Meta, among others, that they have boasted to their advertisers that they know pretty precisely when a teenager feels ugly, stupid, and worthless – and that this is the perfect time to expose them to a beauty ad. One can imagine the same thing happening to people in mourning.
These technologies are found in a poorly regulated market for AI consumer tech. Here, data ethics and data security are generally non-existent. This in itself poses a major risk.
”Who did you Love Most, Dad”?
While one thing is the role that the ad tech industry can play here, another is the fact that chatbots hallucinate and can be trained to weight certain things depending on the provider’s values. This can give a lot of power to those who help develop the app. They can—perhaps unintentionally—help create a reality and an image of the deceased and themselves that matches their understanding of the world. For example, what happens if you ask, “Who did you love most, Dad…?”
People are extra vulnerable when they are grieving. They need extra protection from companies that claim to be able to fix something that feels very intense and uncomfortable – but that perhaps does not need to be ‘fixed’…
In a video featuring CEO and founder Justin Harrison of You – only virtual, a company that develops grief bots, he states: “What I would like to see is the complete and total eradication of grief. The feeling of grief that comes with losing people.”
If we attempt to eliminate grief, we may also be eliminating something deeply human.
In the interview, Harrison is asked by the journalist if he is lonely. He says that he does not think he is any lonelier than others – he talks to many people – but like many others, it is mostly with people he does not know that he chats with online.
And perhaps the fact that more and more people prefer to share personal things with AI chatbots rather than with people while they are alive (because chatbots ‘do not judge’) is the perfect primer for the growth of grief tech in society.
But perhaps also for a reification of the deeper layers of existence?
By sharing our innermost feelings with machines that themselves have no feelings, we are participating in a huge experiment. For how does it affect us and our humanity – on both a large and small scale? As individuals and as a society? More and more people are reporting loneliness, and Zuckerberg and Altman’s answer to that is to sell some more AI companions. AI will solve it. Problem fixed.
I strongly disagree with them. The answer to loneliness is not products – but companionship or community. But social interaction is complicated, it can hurt, and it takes practice. As a species, we are completely dependent on figuring this out.
The Complexity of the Real World
But the real world tends to be less black and white. Perhaps not all bereaved people want the illusion that the deceased is still alive?
Perhaps the app may even be difficult to get rid of. Is it designed with a respectful option for a final digital farewell? Are we passing on a responsibility to our descendants that many of them may not want at all? Give the dead peace. And the living?
I have spoken to many people who compare the grief bot to, for example, a psychologist or other therapist. But they do not pretend to be like the deceased, do they? And they have to comply with a number of ethical and legal codes. But with these products, tech companies can largely do more or less as they please.
More Human Care – Less Silicon Valley
For all of the above reasons – and more – I am very concerned about this type of grief tech based on GenAI. Not because I don’t recognize that there are as many ways to grieve as there are people who have lost someone.
But I would insist that grief is neither a problem to be fixed by tech or new business opportunities, nor a phenomenon to be eradicated. It is part of life, and all of us around the bereaved must do our utmost to be there with warm hearts and hands and a willingness to surround them at a time in their lives when they may be at their most vulnerable. We generally need more humanity – not more interaction with machines. Especially – but not only – in the vulnerable periods of life.
Photo: From the science fiction TV series Black Mirror, the episode Be Right Back from 2013, where a woman revives her boyfriend, first as a chatbot, then as a chatbot with a picture, and finally he comes back as a silicone robot. In 2013, this was truly science fiction. Today, we have already taken the first two steps with services that can ‘revive’ people as digital avatars. We have yet to see them come back as physical robots.
Translated with support from deepl
