Until technology allows us to upload our consciousness to a computer when our physical bodies start irreparably failing, death is going to remain a real thing. But what if you could continue communicating with loved ones — or, at least, a reasonable facsimile of them — long after they’ve shuffled off this mortal coil? It might sound like an episode of Black Mirror (it is!), but it’s also the basis for a recently announced research project being carried out at India’s Shree Devi Institute of Technology.
Researchers Shriya Devadiga and Bhakthi Shetty have been investigating how a chatbot could be made to duplicate a person’s personality digitally, granting users the ability to chat with an A.I. approximation of an individual, such as a family member, who is no longer around.
For their study, the researchers used Replika A.I., an app created by Russian coder Euginia Kuyda. Replika trains a chatbot designed to replicate an individual’s communication patterns by using their digital conversations as training data. Through pattern matching, the more you chat to your Replika A.I. chatbot, the more its sentences sound like something that you would say. Or, in the case of Devadiga and Shetty’s proposal, something that your deceased relative, loved one or friend might say.
This could be achieved by feeding it the sum total of an individual’s available social media presence, tweets, emails, and any other relevant information, to produce a virtual entity that is as close to indistinguishable from them as possible. Think of it like a Turing Test with a touch of Frankenstein thrown in for good measure.
“[Something such as this could help] people to overcome their trauma after losing their beloveds, is what this use case serves,” Shetty told Digital Trends. “In a fast growing world full of technologies — and especially a booming [field like] artificial intelligence — I think this would prove desirable. Because in such a world, nobody would like to spend most of their time depressed or mourning over their loss Now this could be chilling, but that’s the truth.”
A research paper describing the Shree Devi Institute of Technology project was recently published in the Asian Journal of Convergence of Technology. Unfortunately, as fascinating as the concept of virtual immortality might sound, at present the project remains a hypothetical one; a peer-reviewed version of the kind of conversation a couple of stoned roboticists might have at 2 o’clock in the morning.
This isn’t the first time that the idea has been brought up, however — with varying degrees of complexity. Given the somewhat unsettling nature of the idea, a number of companies have tried to tap into this nascent netherworld market. This ranges from efforts like Google’s Inactive Account Manager (“Make a plan for your Google Account if you pass away”) and the startup DeadSocial, which simply allow users to bequeath their data to people they trust after their death, to more fully realized attempts to use this data to do something frequently unsettling.
Sadly, not all of these latter attempts have been quite so successful. The startup LivesOn burst onto the scene in 2013 with its promise to use machine learning to sift through your past social media feeds to comment on Twitter news stories long after your expiration. Its brilliant tagline? “When your heart stops beating, you’ll keep tweeting.” Sadly, LivesOn no longer does — and its website now redirects to a placeholder lifestyle website.
The company Eterni.me promises something even more ambitious, the chance to have your personality encoded as an interactive avatar that can communicate with friends and family. However, it is still in private beta — although the 42,116 people who have signed up suggests that there is at least some degree of interest.
Perhaps the most ambitious attempts to encode personality into a machine are the efforts which include robots. For much of the past decade, American entrepreneur Martine Rothblatt has been gradually developing BINA48, a head and shoulders bust robot modelled after her real life wife. While the real Bina is alive, the goal is nonetheless similar: to create what Rothblatt calls a “mindfile” of her spouse that is indistinguishable from the real thing. Rather than being trained only on social media posts and other data not designed for this purpose, BINA48 is explicitly programmed with information about Bina’s memories, feelings, and beliefs. The resulting attention-grabbing robot has enjoyed countless TV and magazine appearances and even, earlier this year, rang the opening bell at the New York Stock Exchange.
On some level, there is nothing niche about the idea that humans would want to use technology to speak with those who are no longer around. When we watch a movie starring an actor who is no longer with us, or read a book by a deceased author, we get a glimpse at a person — or even into a mind — that is no longer counted among the living. Applying A.I. to this could simply add a level of interactivity previously unavailable to us.
Back in the 1980s, artificial intelligence concerned itself with the so-called “expert system.” This was based on the idea that specialized knowledge could be transformed into a series of probabilistic rules. If they had lived up to their potential, it would have been only logical to try and reproduce and distribute the thought processes of the world’s most brilliant CEOs, physicists, artists and others so that they could continue advising or creating from beyond the grave.
Personally, I’m not sure that the idea of giving someone a chatbot version of a deceased parent or spouse will ever be the same as the real thing. Suggesting so diminishes the most meaningful relationships we enjoy in the real world. But at a time when more people are chatting with A.I. chatterbots like Microsoft’s Xiaoice than ever, perhaps this is simply an impulse from a bygone age. Thanks to advances in robotics, “deepfakes,” machine learning-aided voice replication and, of course, the enormous gobs of personal data produced each day, the idea of creating a passable avatar version of a dead loved one is no longer entirely science fiction. It just needs a company with deep enough pockets to come along.
“I see the value in something like this from a more historical context,” James Norris, the creator of DeadSocial and founder of the Digital Legacy Association, told Digital Trends. “If you could ask your great grandfather what it was like during World War II or similar, those kind of automatic chatbot responses could be interesting from a family history perspective — whether it’s in physical form like BINA48 or only digitally. From a bereavement perspective, I think it’s harder. I can foresee there being potential problems from a safeguarding perspective. You would want to work with a bereavement counsellor and bereavement charities [to make sure this was done properly]. There are so many implications that are unknown right now.”
Or, as Jeff Goldblum’s character so eloquently put it in Jurassic Park: “Your scientists were so preoccupied with whether or not they could, they didn’t stop to think if they should.”