In the future there will be mindclones

This an exciting excerpt from the new book Thinking Machines by Like Dormehl. The book offers a detailed history of primitive machine learning and a dense and fascinating look at the future of true artificial intelligence. The book is available now.

Marius Ursache wants you to live forever.

It’s not a shock to find out that, in an industry that skews as young as tech, few people spend much time thinking about death. This is, after all, a walk of life in which twenty-one- year-olds are already onto their second startup, billionaires are minted by twenty- five and even Steve Jobs once fretted about whether people older than thirty were capable of achieving anything of lasting significance.

As a result, the idea of growing old and dying is, for most Silicon Valley denizens, the furthest thing from their mind. As a former medic from Romania, Ursache thinks about death more than most people. He has even turned it into a job. As the creator and founder of a startup called, he spends his days working toward the dream of building Artificially Intelligent 3- D avatars: digital beings that will look, sound and, most important of all, act like individuals who are no longer with us. Ursache’s journey began several years ago when he became fascinated by the game Second Life, a vast online virtual world created by the San Francisco- based developers Linden Labs. Although Second Life resembles a computer game, it differs in one crucial sense. Rather than featuring set objectives and manufactured storylines, players in Second Life refer to themselves as “residents” of the game, and participate in any way that they wish, whether that means running a shop, or simply hanging out with friends.

“One day, I started wondering what happens to a person’s avatar in the game after they die,” Ursache says. Was there, he pondered, a kind of Second Life purgatory where abandoned avatars lived on in a zombie-like state, long after their human operators had passed away? What would happen if one tried to interact with these avatars?

He began to consider the idea more and more. He attempted to work out the logistics of programming an artificial agent that could convincingly mimic the behavior of its human counterpart. He thought about the kind of code one would need to write for an avatar so that it could learn to move the way its human player once moved, to talk the way that they once talked, and to form and pursue the kind of goals they might have created and pursued. And as with any enterprising entrepreneur, he tried to think of a way to turn it into an actual product.

In February 2014, Ursache was invited to attend a program for entrepreneurs at MIT, by a mentor he had met in Bucharest. As part of the program, he was asked to come up with an idea for a project to work on. By this time, the idea had expanded in his mind. “What I was thinking was that this could be a great way of letting you collect and curate your digital footprint throughout your life,” he says. “The avatar would be an interface for accessing that information.”

He pitched the idea to the group as “Skyping with dead people,” and hurried to note that a lot of the AI technology needed to bring such a project to life already existed in various labs around the world. Despite the group receiving a total of 130 ideas— of which Ursache acknowledges his was the oddest— “Skyping with dead people”was accepted as a project worth pursuing.

Ursache had his reservations, however. “I knew that it would have to do something more than just simulating a conversation with a dead person,” he says. “That would be too fucked up. It would mess with the grieving process and, frankly, would just be weird.”

He decided to put up a webpage to gauge the reaction of the general public. If they responded positively, he would keep working on it. If the idea met a wall of indifference, or even anger, he would drop it.

Within the first four days, the page had 3,000 people sign up to register their interest. That number quickly rose to 22,000, and then kept right on climbing. There were plenty of messages, too, which Ursache dutifully read as a form of market research. Most of them were full of praise for the project, although a certain percentage (he estimates around one- fifth) talked about how creepy it all sounded. Who wanted a version of Siri that acted and spoke like their dead grandparent?

Then Ursache received the e‑mail that changed his life. It was from a person dying of terminal cancer. In their e‑mail they explained that they had six months left to live. A project like, they wrote, was their chance to leave something behind for friends and family.

“It was easy to reply to the messages from people who were congratulating or criticizing us,” Ursache says. “But what could I say to someone who was dying? That was the moment I decided that this was something worth dedicating my life to.” Almost overnight, Ursache made the decision to pack in his previous job and focus on full- time.

Today, has 30,269 eager subscribers, all waiting on their ticket to digital immortality. The company’s website shows video clips depicting a range of memories from the average lifetime. A bride and groom kiss on their wedding day. A mother hugs her baby. A child plays at being a superhero in the garden. Graduating students throw their caps in the air. Retirees laugh together. “What if . . . you could preserve your parents’ memories forever?” Ursache’s marketing blurb reads. “And you could keep their stories alive, for your children, grandchildren and for many generations to come? What if . . . you could preserve your legacy for the future? And in this way your children, friends, or even total strangers from a distant future will remember you in a hundred years? What if . . . you could live on forever as a digital avatar? And people in the future could actually interact with your memories, stories and ideas, almost as if they were talking to you? collects your thoughts, stories and memories, curates them and creates an intelligent avatar that looks like you. This avatar will live forever and allow other people in the future to access your memories.”

Currently the technology doesn’t exist to allow us to “Skype with dead people” as Ursache would eventually like. While his team work on the machine- learning tools that will make the technology a reality, instead focuses on collecting the users’ data that will one day give its avatars their digital lifeblood. He doesn’t think’s 30,269 early adopters are going to be waiting forever, though.

“This isn’t technology that is decades away,” he says. “Building lifelike avatars is an iterative process. Think of it like search results; they’ll just get better and better, more and more accurate as time goes on.”

Excerpted from THINKING MACHINES: The Quest for Artificial Intelligence—and Where It’s Taking Us Next by Luke Dormehl, published by TarcherPerigee, an imprint of Penguin Publishing Group, a division of Penguin Random House LLC. Copyright © 2017 by Luke Dormehl.