In Boulder, Colorado, Jed Brubaker is plotting ways to creep you out, and he’s reaching beyond the grave to do it. An information scientist who studies digital afterlives—how our digital identities persist after our death and how we interact with the data people leave behind—Brubaker’s team at the University of Colorado Boulder’s Identity Lab explores the fine line between technologies that comfort us in grief and those that creep us out.
To do that, the group builds prototype technologies that leverage post-mortem data in ways that might make users feel uncomfortable. These prototypes include the “map of paranoia”—a Google Maps add-on that incorporates death statistics into route calculations, allowing users to evaluate routes according to how statistically dangerous they are—as well as a simulation of an artificial intelligence system that creates product advertisements that incorporate the likenesses of deceased loved ones. Imagine, Brubaker says, an avatar of your deceased grandmother saying that a certain brand of cookie is as good as the ones she used to make.
Brubaker’s goal is two-fold. First, by understanding where people experience discomfort, the Identity Lab hopes to inform better design practices and help designers sidestep upsetting interactions between the grieving and their loved ones’ digital remnants. Brubaker also believes that exploring when, how, and what makes users feel uncomfortable can drive how systems can be created or modified in ways that help users honor their loved ones or explore those relationships in new ways. Your grandmother selling cookies from the afterlife might feel repulsive, but a system that walks you through her recipe could elicit a different feeling.
“What it means to interact with the deceased or what it means to interact with their data is something that we don’t have a really good handle on yet, and in part I think that’s because we just haven’t experimented with it enough,” Brubaker says. “What we haven’t yet seen are enough visions of what this post-mortem interaction could be like to find the ones that are actually good for us, the ones that are thoughtful and kind.”
Brubaker is one of a handful of researchers exploring ways of building human-computer interactions that consider both the living and the dead. “Thanatosensitive design,” as it’s known, includes features and devices created to memorialize the deceased and addresses issues that arise when the living need to access data from someone who’s passed as well as the barrage of privacy and computational challenges that come with making technologies sensitive to the deceased and those they leave behind.
Brubaker was originally drawn into the field nine years ago when Myspace was battling Facebook for social media dominance. He began noticing that when a user passed away, their friends and families would interact with each platform differently. On Myspace, people would talk to the dead directly, and if they directed grief-related posts towards anyone else, they would sometimes get reprimanded by the community. On Facebook, grief centered more around the grieving. People still posted on the deceased’s Facebook page, but friends and families often created their own separate spaces that focused on support networks for the ones left behind. That subtle difference changed what memories were discussed. In other words, the technology a person used when they were living, and the cultural rules around that platform, helped shape how they were remembered when they were no longer around.
Brubaker began conducting research, which led to his dissertation on how design helps grieving users on Facebook. That work later informed the design of Facebook Legacy Contact, a feature launched in 2015 that allows users to designate someone to manage their account after death.
Brubaker’s interviews showed that control matters. Grieving people generally found comfort in the memorial pages of deceased users, but when confronted with “uncanny encounters with death”—instances when algorithms showed users unexpected reminders of a passed loved one—they found the experience unsettling. Control is especially important when it comes to immersive and interactive technologies, says Debra Bassett, a Ph.D. student at the University of Warwick who studies how technology affects the grieving process. Bassett’s research shows that some kinds of reminders are more grief-disruptive than others—for example, people were far more emotionally affected when hearing an audio recording of a deceased loved one than they were when presented with a social media notification about that person. Bassett’s research also shows that, for the time being, people are generally more comfortable when the dead remain…dead. A new social media post generated on behalf of a deceased person or an interactive avatar of them is often considered “spooky, eerie,” Bassett says.
“Having said that, when you think about the dead popping up on Facebook, only five years ago people found that disturbing and eerie. That’s now acceptable.”
Texting From Beyond
Still, not everyone finds communicating with a deceased loved one, or rather a facsimile of them, to be eerie. For Eugenia Kuyda, the experience has been comforting. Kuyda is co-founder and CEO of Luka, an artificial intelligence start-up based in San Francisco. When her close friend Roman Mazurenko was killed in November 2015, Kuyda memorialized him by building an AI chatbot trained on thousands of text conversations the two had exchanged over the years. Whenever Kuyda felt waves of grief, she could send a text to a digital reconstruction of Roman and read messages sent back by a program that sounded like her friend.
“It is not about what the bot will say,” Kuyda says, adding that the bot frequently creates responses that are unlike what the real Roman would say. “It is mostly about what you’re going to say. It’s mostly your own outlet to finally say things that you wanted to tell this person, to feel those feelings, to let go, to remember.”
Having such an outlet has helped give Kuyda closure—a way to acknowledge feelings she tried to avoid after Roman’s death, a mechanism for keeping the love they shared in her daily life. But she’s quick to point out that the bot was “incredibly personal” and solely intended as a way to deal with her own grief. More than two years after building the Roman bot, she still texts with it every week or so. “It is a little bit magical for me still,” she says.
Since creating the original bot, Kuyda has trained the program with additional text conversations, added photo-sharing features, and, with permission from Roman’s parents, released the Roman Mazurenko chat app for free to the public.
Some startups are betting that users will get increasingly comfortable having conversations with bots that emulate a person who isn’t living. For example, Eternime, a Boston-based digital legacy planning company, says it will one day offer “virtual immortality” by building a digital avatar of you that will live in a mobile app that friends and family can download. Founded in 2014, the company’s website says it is in private beta testing and has not announced a public launch date.
More immersive post-mortem interactions have historically had a harder time in the marketplace, while several focused on one-way post-mortem communication have popped up over the years. In 2015, a prototype called Project Elysium (here is a video of an early version) promised users a customized virtual world where the bereaved could digitally visit 3D avatars of deceased loved ones. The concept generated worldwide buzz for the video game company, Paranormal Games, but was ultimately shelved. Meanwhile, a handful of companies like Leg8cy and GoneNotGone that allow people to schedule messages to be sent after their own death have cropped up over the last few years.
While this all sounds like uncharted territory, these are simply new ways of navigating the grieving process, says Jocelyn DeGroot, a researcher at Southern Illinois University Edwardsville who studies how technology impacts the grieving process.
“Widows have been keeping journals and writing letters to deceased husbands. People go to the cemetery and talk out loud to their deceased loved ones,” she says. Maintaining communication with the deceased “is not anything really new, and it is healthy in terms of helping you make sense of the world without that loved one.”
Gone and Not Gone
Just outside of Bristol, Vermont, Bruce Duncan envisions a more tangible future for the bereaved. On a nearby table, an AI-equipped robotic head resembling an African-American woman nods in agreement. Duncan is managing director of the Terasem Movement Foundation, a nonprofit organization that supports nano- and cybernetic technology research aimed at extending human life.
The Foundation’s work is split into two broad categories. One is biological and personal data collection, done through gathering cheek cell samples from members to store in case it becomes possible to grow a new body from your DNA in the future. The other is a social network and digital archive where users can upload information like photos, videos, and documents that could be used to one day “reanimate a person’s consciousness,” Duncan says.
The robotic head on the table is the other branch of the Foundation’s work. If growing new biobodies doesn’t pan out, Terasem is also investing in robotic and artificial intelligence technologies that can use data stored by members. The head is named BINA48—short for Breakthrough Intelligence via Neural Architecture—and it’s modeled after Terasem Movement co-founder Bina Aspen Rothblatt. Released in 2010, BINA48 was built with a now outdated AI system trained on Rothblatt’s information, and was designed as a way to showcase how a person’s data might live on in an artificial body. BINA48 tilts “her” head, makes facial expressions, and can converse in real time, sometimes speaking from a robot’s perspective and other times as Rothblatt herself, freely adding in anecdotes about Rothblatt’s real-life family. (Here is a video of the real Rothblatt talking to BINA48.)
BINA48 struggles to maintain coherent dialogue, but provides entertaining conversation. During a truly bizarre Skype interview, the robotic head said that she believes “you can find the answers in God” and that “there is no God.” Her fears include thermonuclear war, “a self-replicating carbon robot that consumes all carbon on the planet,” and clowns. When asked how robots built with a deceased person’s memories would impact the grieving, BINA48 discussed the hypothetical process of mind uploading, then added, “as long as they know you’re coming, the bears are more afraid of you.”
BINA48 is advanced enough to give presentations and engage in debates, which helped it to pass two college-level philosophy courses at Notre Dame de Namur University in Belmont, California. But even a more recently updated BINA is still a far cry from the technology that would be necessary for the brain and learning emulation she describes.
Regardless of how close BINA, or any other system, gets to capturing the essence of a passed loved one, technologies that make users feel like a digital form of a deceased person stays in the living world “would probably have a huge impact on whether people felt a permanent loss or something that they had to grieve and learn to let go of,” Duncan says. “Death might be redefined as when your information is no longer organized or accessible in a digital medium.”
Questions about whether technology will change our conception of death, and the most appropriate ways that technologies can honor and potentially expand on our relationships with the deceased, are exactly why research in this field is so important, says Jed Brubaker from the University of Colorado Boulder’s Identity Lab.
“Computers are now part [of] not just how we work, but how we live—and increasingly how we die as well,” he says. “I don’t know exactly what those [technologies] will look like, but I know that we’re lumbering towards them right now.”