The theory that humans will eventually be able to upload our brains to computers has fascinated futurists and neuroscientists for years. By transferring our minds into machines we could live forever, unmoored from the feebleness of our physical bodies. The concepts of death and bereavement as we know them now would cease to exist.
Currently the idea lives within whitepapers and sci-fi movies, and the only thing (most) researchers agree on is that it won’t be possible for a really long time. But while we’re far from achieving that pinnacle of immorality, technology in the here-and-now has already started giving us a sliver of eternal life while shaping how we grieve our loved ones if and when they die.
The Silicon Valley-startup Eterni.me aims to let people preserve their “most important thoughts, stories, and memories” in an artificially intelligent system that could ultimately communicate conversationally with others once its creator is gone. Such an avatar could both soothe the bereaved and allow future generations to know much more about their great-great-grandparents.
Since the company had a splashy launch two years ago it has progressed with crypt-like quietness. Their site is mostly unchanged, save for the counter of people signed up for early access that rises each day. But now it says it plans to launch revamped version of its app to a test group in early 2017. The app will connect to a person’s Facebook account (with other social integration planned) and launch micro-conversations based on their daily activities. Ultimately that process will build the AI’s understanding of its subject, and be able to function like a smarter, more personalized version of the chatbots you can already interact with today.
“We chose this approach because we would like to help people solve the ‘blank page syndrome’ we have often seen when somebody tries to write in their diary,” explains Eterni.me founder Marius Ursache. “We want to encourage people to start curating their digital legacy and we will do that by triggering conversations based on previously collected data of what they did, felt, or experienced.”
Though Eterni.me’s new app will be collecting all this information, the neural network backbone won’t be able to turn that mass of unruly data into engaging conversations yet.
“The artificial intelligence that we need requires a lot of effort and time — it may take years,” Ursache says. “A lot of the people who expressed interest in Eterni.me don’t have this time. This forces us to use more or less the the approach of the cryonics initiatives — preservation first, technology second.”
Meanwhile, a semblance of the company’s vision has already come to “life.”
Earlier this year, entrepreneur Eugenia Kuyda created an AI-powered chatbot in honor of her friend Roman Mazurenko, who died tragically in 2015. Kuyda and other mourners could “speak” with Roman by talking to the bot or asking it questions — its responses would be based on the thousands of texts messages and photos she had fed the neural network controlling its artificial intelligence.
While the Roman bot couldn't narrate stories in the same way that Eterni.me envisions, it wrote in his voice. Talking to the bot helped Kuyda grieve and made her feel more connected to her friend.
Inspired by the personal conversations that other people had with Roman’s bot, she’s now working on Replika, a new product that, like Eterni.me, will collect a person’s stories and memories and learn to mimic their conversational mannerisms by texting with them. The bot would first ask you about the past and eventually prod you about your daily life. Your friends could talk to your Replika, too, asking it personal questions or using it to schedule plans with the real you.
Get the mach newsletter.
“It's all about making people a little bit more open about themselves ... a little more vulnerable,” Kuyda says. “Now even the first Replikas we have on the app can tell you a lot about the person that trained it.”
The goal of the platform, she explains, isn’t to let everyone create their own “Roman” that will live on after they die, but to give everyone a kind of “friend” that is there every day to talk to, to make you “more connected with your friends and yourself.”
But intention or not, a person’s Replika bot would be able to keep responding like you and telling your stories after you died in the same way that Eterni.me’s avatars could.
"Imagine if someone had an extensive, almost inexhaustible archive of conversation available with the deceased."
Over 100,000 people have reserved their Replika “name” and 17,000 applied for early testing of the app, but only 1,200 were given access to the beta version. Kuyda was vague about when she thinks the technology will actually be good enough for the general public.
The natural progression of Eterni.me or Replika’s ideas would be to combine the AI with realistic digital 3D avatars (something Eterni.me eventually plans to do to, but has back-burnered while it works on the first part of its mission). Facebook is researching how to create life-like virtual reality avatars as are a host of different startups, including one called XXarray that explicitly envisions helping people create a digital legacy.
Long-time VR researcher Jacki Morie coined the phrase “the ultimate selfie” to describe how she predicts that current technologies could eventually combine to create digital avatars that would not only would look like us, but move like us, and learn from us, too.
She imagines that in the not-so-distant future, when using an avatar in VR is fairly common practice, people will be able to train a digital surrogate that could live on after they’re gone.
“You’ll be scanned and have this 3D construct that can then walk around in virtual worlds and it will retain the things that you did within an AI,” she says. “This kind of avatar could become a repository for our experiences and our memories.”
While the idea of talking to a dead relative through a digitized ghost may seem creepy — edging into bizarre “uncanny valley” territory — experts say that, on principle, it’s actually a completely natural progression of the way that people have always grieved.
Heather Servaty-Seib, a grief researcher and Purdue professor says that it’s in line with the “continuing bonds” theory of bereavement. Death changes the form of your relationship with someone, not the fact that you have one.
Many already talk to the dead through a smartphone screen instead of at a graveyard. (Or through a phone booth, like in Otsuchi, Japan where tsunami survivors “call” their lost loved ones.) Facebook allows people to “memorialize” the pages of the deceased and those pages often become a public and wide-reaching digital wake where friends and family share photos, memories, or their messages to the dead.
AfterTalk prompts the bereaved to write letters to their loved ones that can either be private or shared with a therapist or family member. DeadSocial allows users to set up a curated Facebook-page of sorts where their love ones can commune once they’ve passed. They can even choose to schedule personal messages that will send from beyond the grave.
If people have always found ways now to talk to the dead, could it be problematic if they start talking back?
Both Servaty-Seib and grief therapist Dr. Bob Neymeyer agree that new ways to communicate with the dead could be healthy for many patients. Still, there are some potential pratfalls related to how we mourn. It could be damaging to children, who generally don’t fully grasp the concept of death, Servaty-Seib says. Or it could be addictive to people already prone to focusing too much on the dead, deemphasizing other relationships and not moving forward, says Neymeyer. He’s had patients who have listened to the dead’s voicemail over and over or spent days on end at their grave.
“Now imagine if someone had an extensive, almost inexhaustible archive of conversation available with the deceased,” he says.
Beyond the psychological implications, palliative care doctor Mark Taubert says that he doesn’t think that it will be technically feasible for the tech to answer the kinds of burning questions people will have for their digitized loved ones.
“I often find that the questions a grieving friend or relative may have for the deceased are often quite difficult to answer, for example, ‘Did you ever forgive me for what I did in ....?’” he says. How could an artificially intelligent system taught simply to mimic your style and echo your memories tackle that difficult question in any real way?
Digital legacy tools like Eterni.me or DeathLogic don’t just serve the people left behind. They are very, very much about the person who is dying or thinking about their own eventual death.
These services offer a way for us to build monuments to ourselves. They let us combine our curated memories and experiences with all our messy data — the texts and tweets and “Likes” and photos — and organize it into something that we hope will let our loved ones and future generations actually understand something about who we were.
Until we can really can upload our brains to the cloud, that’s our best shot at immortality.