Stayin’ alive, staying’ alive: Echoes of you will likely linger in a digital afterlife
Startups are already resurrecting the departed as chatbots. A digital you could end up ‘reading to’ a grandchild you never met.
There’s at least one kind of afterlife we can be certain of: the digital one.
We are all going to live on, in some form. What this means varies wildly, depending on whom one speaks to. For the average user of the internet, the idea of this afterlife is limited to search histories, social-media profiles, and possible memorial pages.
To the tech innovator, the answer is literally out of the dystopian near-future series Black Mirror. There, a person could be reanimated, with an android lookalike taught to speak as they might, using posts from their social-media feeds.
In the episode Be Right Back from Season 2, a pregnant woman loses her partner, and a friend orders her a replica in an attempt to help ease her grief. It starts out as an entity that texts in what could reasonably be the dead man’s voice. Then, the grieving woman requests the company to upgrade her plan to enable phone calls.
Upon further payment, a package arrives at her door. It’s a new him — one that cooks but doesn’t eat, seems to know her well, but has nothing new to offer. He can never say anything to her that he hasn’t already said; he also cannot register most emotions.
Eventually unnerved, she decides to distance herself from the android. And so it is that her child grows up thinking of the being in the attic as a friend to be visited on special occasions, like her birthday.
While this is a uniquely disturbing take, we are some way down an eerily similar road.
Today’s inventors are viewing the afterlife through two lenses: ways to delay death, and ways to use artificial intelligence (AI) and virtual reality (VR) to keep a person who has died notionally alive.
OpenAI for longevity
The med-tech start-up Retro Biosciences, based in San Francisco, has a stated mission to “add 10 years to a healthy human lifespan”. Methods being explored to stall ageing, and perhaps reverse it, include cellular reprogramming (the idea that one can return an adult cell to a stem-cell-like stage, in a loop, thereby stalling damage and decay), autophagy (the process of reusing dead or damaged cells; something the body already does in a limited way) and plasma-inspired therapeutics (the rather scientifically shaky idea that infusions of younger blood can delay cell damage).
In March, OpenAI CEO Sam Altman announced that he was investing $180 million in Retro Biosciences. He called it “OpenAI for longevity”.
The company has said in statements that the molecules for its autophagy programme will likely enter the clinical-trials phase in about a year. The plasma programme, in two years.
The cellular reprogramming effort, it has said, holds out the most promise, but is farthest from proof-of-concept. “We will work towards a clinical proof-of-concept over the next four years,” the company said in a statement earlier this year.
Still here?
And then there is the virtual hereafter.
Eugenia Kuyda talks to her dead friend, Roman Mazurenko, regularly. He replies to every message. She tells him things she wouldn’t have told him when he was alive, she told The Guardian in an interview in 2018.
Kuyda is the founder of Replika, a Russian start-up that uses an AI-driven neural network (so called because these programs are modelled on the human brain’s rapid-fire input and response mechanisms) to resurrect a version of a dead person as a chatbot.
Mazurenko was Replika’s first project.
To create the virtual profile of her friend, who died in a car accident, she asked his friends and family to share old text messages they had exchanged with him. She fed the messages into her program, to create a “personality” and determine the tone of the chatbot’s replies.
There is an option to create an avatar too, with customisable hairstyles and outfits.
The Replika app is available for Android and iOS, and on the Meta virtual-reality platform Oculus. The company says it has had 10 million downloads.
A free version lets users chat with a default chatbot named Hope.
It bears mentioning here that there is very little in terms of legal frameworks globally to deal with the implications of a digital afterlife. Meanwhile, the number of services and the range of offerings is growing.
HereAfter AI, launched in 2022, allows people to create a digital legacy of their own. Users can set up an account, take a questionnaire (it covers childhood memories, family, relationships, like and dislike, personality traits, and career details, among other things), add photos, and eventually invite friends and family to access this archived version of who they were. The narrator is the user; visitors revisit memories guided by the person who has presumably departed (it would be exceedingly odd, even in our rather odd world, to invite people in while one was alive).
Start-ups such as SafeBeyond (launched in 2015) and GoneNotGone (2016) also allow people to record messages and videos that can be dispatched to loved ones after their death.
In a world that is already navigating vanishing lines between the real, the virtual and the imagined, the psychological implications of such services are vast.
They could cause more trauma than they ease; could serve to prolong grief; and lead to broader issues of immersion in a world that is essentially a trap of fabricated memory.
“While AI can’t eliminate that pain of loss, it can definitely make the memories last,” is how Alexa senior vice-president Rohit Prasad put it in 2022, speaking during the Amazon re:MARS conference (focused on Machine Learning, Automation, Robotics and Space). He then unveiled a new Alexa feature that allows the virtual assistant to recite content in the voice of a lost relative.
Feed Alexa a short audio recording of grandma reading a book, and the bot can use it to have now-deceased grandma read little Ayesha bedtime stories every night.
They should have used that in Black Mirror.
