Trigger warning: This article mentions themes of death or dying. If at any point the content is distressing, please reach out for support via Griefline or refer to the services listed at the end of this article.
Rumi once wrote, ‘Anything you lose comes round in another form.’ (Goodreads, n.d., p. 1).
There are many ritualistic ways to memorialise the death of a loved one, but what if they had never “died”? Over the past decade, the intersection of technology and mental health has given rise to innovative solutions for various psychological conditions. From virtual reality therapy for Post-traumatic Stress Disorder (Kothgassner et al., 2019) to prescription video games aimed at helping children manage Attention Deficit Hyperactivity Disorder (Tiitto & Lodder, 2017), the mental health technology industry has expanded significantly. Enter, a recent addition to this landscape - the grief bot.
In 2015, Roman Mazurenko, an entrepreneur and prominent figure in Moscow’s night-life scene, suddenly passed away from a fatal car accident (Newton, 2016). His close friend, Eugenia Kuyda, proceeded to create a “digital monument” in his memory (Newton, 2016). While grieving, she found herself re-reading all his old messages, feeling nostalgic at Roman’s unique word choices, and spelling. Kuyda had previously founded a startup involving artificially intelligent chat bots. After the incident, she fed her bot with Roman’s text exchanges. The bot then adopted Roman’s speech pattern, enabling her to chat with a version of him. This marked the birth of the griefbots, or chat bots programmed using digital remains (emails, text messages, social media posts) of a deceased individual to support their grieving loved ones. In other words, using natural language processing, these bots are able to mimic conversational patterns using the data of the deceased.
Are these conversational patterns accurate? How then, does this impact the way we grieve? Should we even be using griefbots? To answer these, we could attempt to understand grief.
Grief is a complex emotion. You could be grieving the loss of a loved one, a relationship, an object, or even an abstract idea (e.g. familiarity). Grief can also manifest at different times for each individual. According to the Australian Psychological Society ‘grief is the natural response to loss and can influence the physical, emotional, cognitive, behavioural and spiritual aspects of our lives.’ (APS, n.d, p. 1). In their book, Elizabeth Kubler-Ross and David Kessel (2014) coined ‘The Five Stages of Grief: denial, anger, bargaining, depression and acceptance. Essentially, the model suggests an initial reaction of symbolic denial or shock. Following this is typically a phase of emotional support through vocalising the experience or making meaning. The final stage being acceptance, or moving forward.
“Are they really gone?”
Denial is viewed as a protective mechanism to meet the psyche where it is.
“Why me?”
Anger is interestingly framed as an anchor to connect you to someone you’ve lost.
“What if they suddenly return?”
Bargaining shifts from the past to the future, until the truth sinks in.
“What’s the point?”
Depression is protecting the nervous system from overload, and is arguably natural to grief, when not clinical.
“I lost them, but I am going to be okay.”
Acceptance as you start to move forward, with some stability.
Now, each of these questions might manifest in different ways, and require different coping mechanisms. However, they do give us an indication of generic phases across unique manifestations of grief. In other words, these are not linear, clear-cut stages, rather, there is an element of individuality in the way we experience each stage. We might experience one stage before another, or circle back, or take a completely new route. In any case, this is one way to make sense of grief. Other theories around grief include Bowlby’s attachment theory (1980) which suggests that our response to losing someone is coded in the way our attachments develop. Silverman and Klass (1996) put forth the idea of continuing bonds, where the meaning of loss changes with the deceased living on in memory. On the other hand, Strobe and Schut (1999) posit a dual process with individuals constantly switching from avoidance or confrontation of loss.
Regardless of your theoretical inclinations, chances are that one might seek closure, a sense of reconciliation or even self-fulfilment after experiencing loss.
What, then, would be the wellbeing impacts of artificial chat bots, that are designed to adopt the language patterns of those we have lost, on the grieving process?
Grief can result in cognitive changes, such as confusion, identity disturbances, dysphoria, and yearning among others (Bonnano & Kaltman, 2001). Norlock (2016) proposes that imaginal relationships with the deceased can reflect relational value, ethical behaviour (such as forgiveness), and relationship maintenance. Furthermore, it is argued that continued internal representations of people who have passed away might also add value to future relationships. In contrast, some may argue that interacting with an artificial grief bot might engender para-social relationships where the user is investing time into a relationship (Vost & Kamp, 2022); however, the receipt is unaware (similar to celebrities and their fans). Furthermore, anthropomorphising a non-living chatbot, and conflating this for a person might distort reality, take wrongful advice, delay grief, or fabricate new false memories (Vost & Kamp, 2022).
It leads one to wonder, just what are the potential ethical issues surrounding griefbots?
Data is impermanent, with the ability to be wiped (Grandinetti et al., 2020). Data is deeply contextual, contingent, and unstable (Grandinetti et al., 2020). In order to understand how the bot is responding, ensuring no advice is given, and also preserving the griever’s best interest, is a complex task. Moreover, viewing griefbots as permanent or true representations of the dead is another issue.
There are also ethical questions around consent and whether the deceased are capable of giving consent to the usage of their data, along with users. Whether companies can be transparent about how the data is being handled, and the algorithms generated, remains unclear. Would knowing how the responses were generated changed the way people viewed grief bots, and would that defeat the purpose?
Yet, there are broader challenges. If users disclose private information to profit-driven companies based on the trust with the person they have lost, the data could be misused. The role of protection plans in the event of deep fakers or hackers, becomes paramount. The large amount of data used also raises questions about the sustainability of such bots. Additionally, the high cost of sophisticated bots might create greater disparities in access to support. While autonomy may improve with access to immediate technology, the addictive interaction patterns could lead to dependence, overuse, and potentially social withdrawal. Furthermore, gender, age, sensitive content, changing political landscapes, might potentially bias the bot inherently.
Griefbots remain a hotly contested topic, with widespread caution surrounding potential impacts. There have been attempts to design similar bots with ethical features in mind, and even suggestions to medically regulate or test such devices. However, this use for AI bots opens up a multitude of questions.
By 2025, Vorst and Kamp (2022) speculate that holographic avatars could be generated through photographs, physical and digital remnants, even voice recordings.
Ultimately, the impact of griefbots on our perception of mortality and memory challenges us to reconsider the boundaries of life, death, and the enduring essence of human connection in a digital age.
Support resources
If you are experiencing prolonged symptoms of grief or depression, please seek support via the following resources with different options for support:
Grief Australia: counselling services, support groups, app
https://www.grief.org.au/ga/ga/Get-Support.aspx?hkey=2876868e-8666-4ed2-a6a5-3d0ee6e86c30
Griefline: free telephone support, community forum and support groups
Better Health Channel: coping strategies, list of support services, education on grief
https://www.betterhealth.vic.gov.au/health/servicesandsupport/grief
Beyond Blue: understanding grief, resources, support, counselling
https://www.beyondblue.org.au/mental-health/grief-and-loss
Lifeline: real stories, techniques & strategies, apps & tools, support guides, interactive
Reach Out Australia: coping strategies
Find a Helpline: for international/country-specific helplines
This list is not exhaustive, please refer to your area’s specific services for additional support.
References
Albert, S., & Bowlby, J. (1982). Attachment and loss: Sadness and depression. Journal of
Marriage and the Family, 44(1), 248. https://doi.org/10.2307/351282
APS. (n.d.). Grief | APS. Australian Psychological Society | APS.
https://psychology.org.au/for-the-public/psychology-topics/grief
Basom, J. (2021, May 19). The ethical, social, and political implications of “Griefbots”.
Bonanno, G. A., & Kaltman, S. (2001). The varieties of grief experience. Clinical Psychology
Review, 21(5), 705-734. https://doi.org/10.1016/s0272-7358(00)00062-3
Craytor, J. K., & Kubler-Ross, E. (1969). On death and dying. The American Journal of
Nursing, 69(12), 2710. https://doi.org/10.2307/3421124
Elder, A. (2019). Conversation from beyond the grave? A Neo‐confucian ethics of chatbots
of the dead. Journal of Applied Philosophy, 37(1), 73-88. https://doi.org/10.1111/japp.12369
Goodreads. (n.d.). A quote by Rumi. Goodreads | Meet your next favorite book.
https://www.goodreads.com/quotes/32062-don-t-grieve-anything-you-lose-comes-round-in-another-form
Grandinetti, J., DeAtley, T., & Bruinsma, J. (2020). The dead speak: Big data and digitally
mediated death. AoIR Selected Papers of Internet Research. https://doi.org/10.5210/spir.v2020i0.11122
Jiménez-Alonso, B., & De Luna, I. B. (2022). Correction to: Griefbots. A new way of
communicating with the dead? Integrative Psychological and Behavioral Science. https://doi.org/10.1007/s12124-022-09687-3
Klass, D. (2021). The sociology of continuing bonds. Culture, Consolation, and Continuing
Bonds in Bereavement, 113-128. https://doi.org/10.4324/9781003243564-11
Kothgassner, O. D., Goreis, A., Kafka, J. X., Van Eickels, R. L., Plener, P. L., & Felnhofer, A.
(2019).Virtual reality exposure therapy for posttraumatic stress disorder (PTSD): A meta-analysis. European Journal of Psychotraumatology, 10(1). https://doi.org/10.1080/20008198.2019.1654782
Kübler-Ross, E., & Kessler, D. (2014). On grief and grieving: Finding the meaning of grief
through the five stages of loss. Simon & Schuster. https://books.google.com.au/books?hl=en&lr=&id=0TltiT8Y9CYC&oi=fnd&pg=PR11&dq=grief+&ots=S1j1XyF91N&sig=pDnxX-bJQIJIFeX074oGrHRD0Ms&redir_esc=y#v=onepage&q=grief&f=false
Lindemann, N. F. (2022). The ethics of ‘Deathbots’. Science and Engineering Ethics, 28(6).
https://doi.org/10.1007/s11948-022-00417-x
Newton, C. (2016, October 6). When her best friend died, she used artificial intelligence to
keep talking to him. TheVerge.com. https://www.theverge.com/a/luka-artificial-intelligence-memorial-roman-mazurenko-bot
Norlock, K. J. (2016). Real (and) imaginal relationships with the dead. The Journal of Value
Inquiry, 51(2), 341-356. https://doi.org/10.1007/s10790-016-9573-6
Santa Clara University. (n.d.). AI, death, and mourning. https://www.scu.edu/ethics/focus-areas/internet-ethics/resources/ai-death-and-mourning/
Schut, M. S. (1999). The dual process model of coping with bereavement: Rationale and
description. Death Studies, 23(3), 197-224. https://doi.org/10.1080/074811899201046
Shardlow, J. (2022). Temporal perspectives and the phenomenology of grief. Review of
Philosophy and Psychology. https://doi.org/10.1007/s13164-022-00659-5
Tiitto, M. V., & Lodder, R. A. (2017). Therapeutic Video Games for Attention Deficit
Hyperactivity Disorder (ADHD). WebmedCentral, 8(11). https://doi.org/10.1101/2020.10.26.355990
Van der Vorst, R., & Kamp, J. M. (2022). 12. Designing a griefbot-for-good. Moral design and
technology, 215-241. https://doi.org/10.3920/978-90-8686-922-0_12