You are using an older browser version. Please use a supported version for the best MSN experience.

Augmented eternity: scientists aim to let us speak from beyond the grave

The Guardian logo The Guardian 27-06-2016 Dan Tynan in San Francisco

undefined© iStock/ Getty Images undefined Would you like a version of yourself to live on after death? A radical new concept called “augmented eternity” could make that fantasy a reality, creating a posthumous impression of our knowledge, opinions and even parts of our personality in digital form.

Researchers at the MIT Media Lab and Ryerson University in Toronto believe that by applying artificial intelligence to all the data we produce each day, we may be able to transfer our thoughts to a virtual entity that not only survives our physical demise but continues to learn as new information is plugged into it. 

Dr Hossein Rahnama, a visiting scholar at the Media Lab and research and innovation director of Ryerson’s Digital Media Zone, is planning to publish a paper on augmented eternity later this year. “My ultimate goal is to bridge the gap between life and death by eternalizing our digital identity,” says Rahnama, who is also founder of Flybits, a cloud-based service that modifies the behavior of mobile apps based on where and how the customer is using it.

“Your physical being may die, but your digital being will continue to evolve with the purpose of helping people and maintaining your legacy as an evolving being.”

It may sound like one of those science fiction brain-in-a-jar movies, but many of the underlying ideas have already been employed in less ambitious ways. For the past few years, researchers have used machine learning and expert systems to analyze large amounts of data, identify patterns and make predictions. It’s what allows Amazon or Netflix to recommend things you might want to buy or shows you’d probably like.

Rahnama wants to apply similar techniques to everything we do. Recorded in digital form, he would then create algorithms to answer questions posed to us from beyond the grave, creating a “new form of inter-generational collective intelligence”.

“In 30 or 40 years we’ll be able to ask questions of scientists or politicians who’ve passed away,” he says. “Imagine that we could activate a profile of Ronald Reagan and ask him what he thinks of Donald Trump.”

The idea that our thoughts will live on after our bodies crumble into dust is not a new one. Ray Kurzweil – father of “the singularity” concept, that man and machine will eventually merge – predicted we’d be able to “upload our minds to a computer” by 2045, but didn’t provide a blueprint for how to do it. Rahnama may have.

Technology is the easy part, he says. But achieving digital immortality will require overcoming some rather daunting challenges around data privacy, as well as a healthy amount of skepticism.

Selecting your digital heir

Thanks to low-cost data storage, ubiquitous internet connections, relatively cheap computing power and increasingly intuitive AI tools, augmented eternity is technically possible today, say Rahnama.

The biggest hurdles: collecting enough data, making it all accessible, and protecting it. The more data you feed an AI system, the more accurate it’s likely to be. To make reliable predictions about what you might have said or thought, if you were still alive, an augmented eternity machine would need vast amounts of highly personal data: your public statements, blog entries, social media posts, photos, texts and emails – even down to what you ate for breakfast (and then promptly tweeted about).

That’s a big reason why Rahnama believes augmented eternity will find its sweet spot with millennials who are already Instagramming, Snapchatting and Facebooking every waking moment.

“You couldn’t do this in the past [because] you didn’t have enough of a digital footprint about someone to be able to do these types of sophisticated predictions,” he says. “Fifty or 60 years from now, millennials will have reached an age where they have collected zettabytes [1 trillion gigabytes] of data individually. What can you do with data? Can you take that data and have enough prediction models to select a digital heir – something that can represent you after you die?”

There are already expert systems that tap the collective wisdom of organizations. The Ross Super Intelligent Attorney uses IBM’s cognitive computer, Watson, to comb through legal documents and provide attorneys with relevant answers to queries about the law.

Rahnama believes it will be 15 to 25 years before augmented eternity becomes mainstream for individuals. The main stumbling block is, who owns the data everyone produces? For his scheme to work, he says, users need to have total control over their data – something companies like Google and Facebook, as well as a multitude of government agencies, might be reluctant to surrender.

“For these services to become mainstream you really need to have privacy organizations to be on board, to introduce this as something that is your right to have,” he says. “The data has to be your own property, which you can pass on as an inheritance, where only you and the people you trust will have access to it.”

In part, Rahnama says, he created the augmented eternity project to encourage research into the questions of data collection and control; to generate policies that deal with the tsunami of data yet to come.

No ghost in the machine

We contacted other AI experts to get their reactions to the augmented eternity concept. Some were cautiously optimistic, others deeply skeptical.

In theory, it should be possible to record everything you’ve said, feed it into an AI model, and have algorithms come up with an approximation of how you might respond to questions, says Van Baker, a research vice-president for Gartner. “This is not off the scale in terms of crazy ideas,” he says.

“The challenge will be nuance. Words are subject to multiple interpretations. Imagine being presented with an outrageous idea and you respond, ‘Sure, we absolutely want to do that.’ That might be total sarcasm, but the algorithm may think you meant it.”

Baker adds that how people respond often depends on context; we might say one thing in the presence of friends, but something much different around strangers or co-workers. And people are often unpredictable or change their minds over time.

“These kinds of things can be very difficult for machines to interpret,” he says. An augmented eternal version of you would probably be at best a shallow replica of your own personality, says Adrian McDermott, senior vice-president for Zendesk, which uses machine learning to determine if customers are happy with their customer support experience.

‘Are we really just a sum of repeating patterns?’ some scientists are skeptical about the augmented eternity concept. Photograph: Alamy

“You could probably build a passable Adrian bot,” he says. “But augmented eternity Adrian wouldn’t be conscious. There is no ghost in the machine. It would remember everything I’ve said better than I could, but it wouldn’t have ideas, wouldn’t be creative. It would be the straw dog version of Adrian.”

Jeremy Pickens, a senior applied research scientist for Catalyst, a maker of electronic discovery tools for law firms, says he’s unexcited by the premise. “I don’t think this is very feasible,” he writes. “The way I understand AI, machine learning and big data is that it works well at distilling large amounts of data into the most common, repeating patterns. And I don’t see the human experience as particularly reducible. Are we really just a sum of repeating patterns? Maybe I’m wrong, but I feel like I’m a different person every year or two … The computer needs to synthesize; it needs to generate new behaviors that it hasn’t actually seen before.”

But not everyone’s a skeptic. Dr Alireza Sadeghian, former chair of the computer science department at Ryerson, says an intelligent system like that proposed by his former student and now colleague, Dr Rahnama, is entirely possible given the advances in computing power and cognitive science.

“If we would have preserved Albert Einstein and then trained the system properly, I think the answers we’d get would be very accurate,” he says. “But I think that using such a system to preserve our own memories would be even better.

“Imagine the ability to have a conversation with a loved one who is no longer with us, the way people still visit Facebook pages of family members who have passed. In terms of both the technology and the service to humanity this would provide – it would be fantastic.”

More from The Guardian

image beaconimage beaconimage beacon