Are they really dead? How is AI is changing the way we look at people who have passed on? Are these crazy questions? Or is technology now able to make people immortal?
Perhaps. Well, almost!
You probably remember watching the Terminator series (If you haven’t watched them, you should!). The story, without giving it all away, focused on Skynet, a massive neural network-based artificial intelligence (AI) system that controlled everything, including the military.
Like the story of HAL in 2001 a Space Odyssey, Skynet became sentient, or self-aware – able to think on its own. It was then a conscious, superintelligence system.
When the humans discovered that it had gained self-awareness, they tried to deactivate it. What followed was Skynet’s decision to retaliate with a nuclear attack to destroy mankind.
Back in the real world, many are concerned about what will happen if AI does become self-aware.
After all, it is used in a lot more than desktops and business applications. AI is also used with ever-increasing complexity as the field of robotics advances, fed in large part by the military. It is an obvious choice to wealthy countries to create self-guided robotic “soldiers,” able to make lightning-fast decisions like who the “bad guy” is and when to shoot, without harming the “good guys,” and without the adversary harming any “good guys.”
Of course, this requires a tremendous amount of input to ensure that correct decisions and actions are able to be performed instantaneously, and without error.
In the old days of computer programming, engineers had the very tedious job of entering binary data, a job that is now fully automated to the point where it is scarcely remembered.
Today, even with a tremendous amount of automation, creating advanced AI takes a huge number of man-hours. Even in the private sector, a seemingly innocuous bit of information left out of a program could be the difference between an autonomous truck crashing into other vehicles, or a large AMR (autonomous mobile robot) seriously injuring a human being on a jobsite.
So, what does this have to do with AI and dead people?
A number of companies are now working on creating AI chatbots to bring people back to life for their loved ones after they have passed on.
What? Wait! Right??
The idea arose out of sorrow. These entrepreneurs have also faced personal tragedy, losing very close friends in tragic car accidents. The death of a father, a friend dying in a a car accident in the U.S. – another dying from being run down by a car in Moscow…. They missed their loved ones terribly, of course, which added inspiration to their ideas.
Marius Ursache, now a fellow at MIT (Massachusetts Institute of Technology), is the co-founder of Eternime (Eterni.Me). Originally from Romania, where he studied to become a doctor, he found his true calling when he switched to technology, taking courses at MIT.
He founded his company in 2014, with the idea of virtually “immortalizing” people by creating a 3D digital avatar that looks and sounds like them after they die. His project became very personal just a short time later, when his best friend died in a car accident.
Enternime was created as an app that collects data about you while you’re alive. It automatically harvests massive amounts of smartphone data – with your permission, of course – and asks you questions via a chatbot.
According to an old Enternime website, “Eternime is a network of Artificial Intelligent avatars that preserve for eternity their owners’ thoughts, stories and memories. The avatar will act as a personal biographer, connecting to its owner’s digital accounts (Facebook, email, calendar) and devices (smartphone, wearables) and collecting data. Various events will trigger chats between the avatar and its owner, making the avatar learn faster. Owners will chat to their avatar almost every day for the rest of their life. Our end goal is to build the largest interactive human-like knowledge and history library, and the most accurate human-like AI entity.”
Their headline on Twitter says that they collect, curate and preserve your memories which are accessible for a digital avatar “who live on for eternity.”
A 2018 tweet from Eternime says that they have been included by the Victoria and Albert Musem in the 100 projects that will shape the future, along with Amazon Echo, Nest, Made in Space, and others.
One beta user who loves the idea, saying that it’s likely that 99% of our memories get lost, so being able to actually leave something behind is an awesome thought.
So what the technology is doing is creating a virtual you, so that when you pass on, your loved ones can still “talk” to you, even long after you’ve left this mortal coil.
Perhaps sadly, Eternime was a great idea but it failed to launch. Its website finally expired on January 29, 2021. The very last post on the Twitter account was on May 9, 2018.
Enter Eugenia Kuyda
Eugenia is from Russia. Until a November day in 2015, her best friend was a young Belarusian man and tech entrepreneur named Roman Mazurenko. He was just over thirty when he was run down by a Jeep on a street in Moscow. By the time Eugenia got to the hospital, he had already passed away.
After his funeral, she reread thousands of messages that she and Roman had exchanged going back to when they met in 2008. They were cherished memories for Eugenia, because Roman had not been a fan of social media and left very little behind to remember him by.
Eugenia, who is also an entrepreneur and software developer, had been creating a messenger app called Luka that was designed to us AI to emulate human dialogue.
After the passing of Roman, she found new inspiration in an episode of the Black Mirror television series, in which a young woman grieving over the death of her boyfriend installed an app that allowed her to continue to chat with him.
People today leave behind a surprisingly massive digital archive of themselves: social media, text messages, digital photos, and the list goes on. Kuyda discovered that they can be very helpful tools for the grieving process.
Eugenia decided to modify Luka to become a real-life version of the app in the TV series. It would give her a way to continue to “chat” with Roman.
She did this by putting his old text messages into the neural network built by her developers at her AI startup. Furthermore, she contacted Mazurenko’s close friends to see if they would be willing to share their text messages with her. Then of his friends and family, including his parents, responded positively to the project, sharing more than 8,000 lines of text.
What Kuyda finally arrived at is Replika, formed on the Luka platform. Her Roman AI had a mixed reception. Some people loved it, finding it an amazing likeness of Roman and a very great comfort to help them through the grieving process. Others did not like it at all. Some even found it creepy. But for Kuyda it was worth it for the comfort it did give.
For her, her continued “conversations” with Roman have brought her a sense of peace. She continues to add material to the Roman bot, and has upgraded the neural network from a selective model to a generative one that can recombine snippets of his texts and recombine them, in theory, to make sentences that are still in his voice.
Replika’s purpose today is to create a personal AI bot “that would help you express and witness yourself by offering a helpful conversation.” It lets people safely share their deepest thoughts, beliefs, experiences, memories and dreams, safely, with a good “listener” who is completely non-judgmental.
The AI can then be passed on to “immortalize” you, should you wish.
James Vlahos is not a scientist. He is a journalist and also the creator of an AI he calls a “Dadbot.”
When his dad was diagnosed with lung cancer in 2016, he started recording all of their conversations for a future commemorative book.
What he ended up with after twelve one and a half hour sessions contained 91,970 words. The offered memories, songs, stories, his dad’s marriage, career milestones, and his driving interests.
When James came across an article about two Google researchers creating a chatbot using movie dialogue on a neural network to create a chatbot that could interact with humans, he got excited.
He decided to use all his recordings to create his Dadbot, a smartphone app that simulated a text conversation with his dad. This has enabled him to continue chatting with his dad, making the loss easier to bear. Even his son uses it sometimes to “chat” with his grandfather.
All this is very interesting, but is it healthy? Does “immortalizing” our dead loved ones ultimately helpful in our grieving process, of does it simply prolong it, giving us a false sense of continuation? Are we playing with fire, edging ever close to creating machines to completely replace human interaction? This is already happening. Artificial “women” are now available to provide company, conversation and much more for men, in what some would find freakishly lifelike capabilities. What if they started thinking for themselves?
The reality is, we really are kidding ourselves and entering into a fantasy world when we start believing AI can really become completely autonomous. But is it okay to kid ourselves that way when we “chat” with loved ones who have passed on? Or is it really any different from looking at old photos and reminiscing?
Perhaps this is just the normal evolution of that very activity. We have gone from memories shared around a crackling fire, to sharing photographs and immortalizing voices on vinyl, magnetic tape, CDs and computers. Perhaps this is merely another point along that evolutionary line!