Virtually Immortal

The advent of technologies like deepfakes and AI-powered chatbots have opened the door to the possibility of a digital afterlife. What does it mean to die when we can still be alive online?

On my 27th birthday¹, I woke up to a news story about Joaquin Oliver, one of the victims of the Parkland shooting.

Pivotal to the story was a PSA video called “The Unfinished Votes,” using Oliver’s likeness and voice, which had been reanimated with artificial intelligence to urge voters to take action against gun violence. The article, and others like it, used phrases like “digital resurrection” and “bring back to life digitally.” A few weeks later, Kim Kardashian West shared a video of a hologram of her late father, Robert Kardashian, with her 67 million Twitter followers. Her husband had commissioned the hologram as a birthday gift to her. The director of Kaleida, the company that produced the hologram, referred to it as a “holographic resurrection.” 

Research from Samsung’s AI lab claimed to be able to create deepfake videos like the Joaquin Oliver ad with a single photograph. The deepfake video used in “The Unfinished Votes” campaign was made with just three photos. It seems now that when we die, we might not stay dead like we used to. 

I saw these videos late in 2020–a year I spent mostly inside, only seeing friends and family through pixelated video chats. Maybe in another year, I would have found it disturbing–being reanimated in a deepfake video, inhabiting technology as a kind of digital ghost. Before 2020, I hadn’t given much thought to the digital afterlife. But then, so many experiences were moved online: school, work, weddings, graduations, funerals. All I saw on the news was an ever-increasing number of people dying from COVID-19–dying, many times, in isolation and forced to say goodbye to their loved ones over Zoom or FaceTime. Death and dying, topics we’ve traditionally shunned talking about in Western society, were suddenly at the front of our cultural consciousness.

We are overdue for some kind of a reckoning with the digital afterlife. For years, Facebook has been running an unintentional side business as a vast online cemetery, with over thirty million dead users. Many of these accounts are memorialized, and announce themselves as “Remembering” the deceased user. But this doesn’t happen automatically: Friends of mine have passed away and I’ve then been prompted to wish them a happy birthday. Immediately following the death of an acquaintance last year, Facebook suggested one of my best friends add her as a contact. Facebook’s one concession to death is its advice to choose a legacy contact who will be able to close or memorialize your account in the event of your death. But again, no one is required to do so. 

What is a body, a face, or a voice, when it is disembodied from the soul that once inhabited it? Do we become the property of those closest to us, like the bodies of brain-dead pregnant women who are kept alive to deliver their children? Can I opt out of being digitally resurrected, or do I opt in? Should I design my posthumous avatar now, to leave some comfort for my friends and family?

The idea of leaving a digital memento is not new, although prior efforts have not been so high-tech. The free “legacy app” RecordMeNow lets individuals record video messages for loved ones, and includes prompts ranging from what makes you happy? to how would you like your loved ones to grieve you? A few years ago, the toy store chain, Build-A-Bear, began offering stuffed animals with custom audio messages that were activated when the toys were hugged or squeezed. Almost immediately, people started using the voice recording technology to either leave a parting gift for their loved ones, or to bring back a piece of their dearly departed. The internet was flooded with videos of weeping adults gripping teddy bears that had the voice of their dead mother, the laugh of their dead spouse. 

But as artificial intelligence has advanced, more options for the digital hereafter have emerged. Eugenia Kuyda, co-founder of the AI start-up, Luka, lost her best friend Roman Mazurenko in a car accident in 2015. Despondent and trying to figure out a way to continue on his legacy, she asked his close friends and family to share their messages with him with her. She then took the 8,000 messages she received and trained a “Roman chatbot” that anyone could write to, and get a message back that would be in Roman’s “voice”. Kuyda made the bot live and shared its link on Facebook. 

Not all of the responses were positive: some of Eugenia and Roman’s mutual friends refused to engage with the bot. Others found it to be a therapeutic, if disconcerting, experience. “There are questions I had never asked him,” one of Roman’s friend’s commented. “But when I asked for advice, I realized he was giving some pretty wise life advice. And that actually helps you get to learn the person deeper than you used to know them.” Kuyda has credited the bot with helping her gain closure on her friend’s death: “I answer a lot of questions for myself about who Roman was.” 

In January 2021, Microsoft filed a patent to create chatbots trained “to converse and interact in the personality of a specific person.” The patent specifies that the person could be dead or alive, and that a user may train a chatbot with their own data, essentially creating their own legacy bot. 

Hossein Rahnama, an MIT Media Lab researcher, entrepreneur, and professor at Ryerson University, imagines a future closer to the Robert Kardashian or Joaquin Oliver version of a digital afterlife. His project, Augmented Eternity, allows you to set up a representative digital avatar to interact with people after your death. Rahnama has stated “his ultimate goal is to bridge the gap between life and death by eternalizing our digital identit[ies].”

The reactions to these digital resurrections tend to fall starkly to one of two sides. Either people find them comforting or horrifying, with little gray area in between. Science fiction, which can help us envisage the future of technology, has explored both sides of the argument.

In 2006, author and neuroscientist David Eagleman published a short story in the international journal, Nature, called “A Brief History of Death Switches,” which begins, “There is no afterlife, but a version of us lives on nonetheless.” Eagleman didn’t imagine our personalities re-animated by hologram or deepfake. He predicted a world in which people programmed complex “death switches” to pretend they were not dead, resulting in “computers operating around the clock sending out the social intercourse of the dead: greetings, condolences, invitations, flirtations, excuses, small talk, inside jokes”. This created an “increasing difficulty in sorting the dead from the living.” But Eagleman’s story ends on a positive note: “This situation allows us to forever revisit shared jokes, to remedy lost opportunities for a kind word, to recall stories about delightful earthly experiences that can no longer be felt…perhaps all that would have happened in an afterlife anyway.” 

The creators of the British science fiction show, Black Mirror, took a markedly different tone when they imagined digital resurrection with a 2013 episode titled “Be Right Back.” A pregnant woman recreates her recently deceased boyfriend using his online communications and social media profiles, even going so far as to give him an identical physical form. At first, she takes comfort in talking to this reproduction. But their relationship goes downhill fast. He doesn’t sleep; he never argues with her. After all, he isn’t her actual boyfriend, just his information and traits stitched together by an artificial intelligence. She ends up locking the avatar in her attic.

The responses to the Joaquin Oliver PSA and the Kardashian hologram were deeply negative. Amongst piles of comments denouncing the videos as creepy and disrespectful and gross, a few people raised the question, How can I make sure no one does this to me after I die? Conversely, when one woman’s Build-a-Bear with a recorded message from her late mother was stolen, it became a viral news story with a $5000 reward offered by Ryan Reynolds (the bear was returned to her). And psychologists weighing in on “griefbots”, such as Kudya’s Roman chatbot, have mostly agreed that the bots offer therapeutic potential, and may help us grieve lost loved ones “a little more easily and a little more accurately than our own memories and objects would.”

It makes perfect sense that people feel so strongly about the uncertainties that these new digital afterlives present. Perhaps the uncanny valley-specific shiver provoked by these avatars is the impetus we need in order to start talking about our digital legacies, because whether we like it or not, we are building those digital legacies online every day. 

Another thorny issue to consider: who will be in control of these avatars once we die? There are few guidelines on how to manage our social media accounts and emails once we pass away, much less account for “the digital afterlife industry,” as Oxford Internet Institute ethics researchers Carl Öhman and Luciano Floridi call it. Some estate planning companies have begun to suggest naming a digital executor in your will — a person to manage and inherit your social media accounts and digital assets after you die.

Öhman and Floridi have a proposal for how to address this dilemma. They suggest treating digital remains as though they were ancient human remains, a situation for which there is a precedent. They argue that “the dead, just like the living, have the right not to be alienated from themselves, a right to human dignity. Just like a human corpse has a right to be treated with dignity, so have our digital remains.” Adoption of this paradigm would require companies offering digital resurrections to adhere to strict regulations about what they could and could not do with the avatar of the deceased person, and might limit the influence of the family of the bereaved as well. It might soothe the worst fears of fans of Black Mirror. 

But law does not evolve as fast as technology, and until such guidelines are enforced, we are on our own when navigating the precarious waters of digital resurrection. We must each decide on which part of ourselves, if any, we want to leave behind.

Footnotes

1. Coincidentally, 27 is a culturally fraught age– the “27 Club,” an unofficial list of musicians and artists who died at the age of 27, includes Jimi Hendrix, Janis Joplin, Jean-Michel Basquiat, Jim Morrison, Kurt Cobain, and Amy Winehouse, amongst many others.

Cara Peralta-Neel (ITP 2020) is a Canadian-American writer and user experience designer living in New York.