Someday, a digital avatar may live on forever that replicates everything about you.

You can already guess at how this might work.

Voice assistants like Apple's Siri and Microsoft's Cortana seem to have a personality, especially when they make snarky jokes.

Now imagine a robot that has your own personality. It can tell the jokes you normally say, use your voice inflection, and possibly even state your political views. This avatar would know, for example, that you're more than a mild Golden State Warriors fan. It could analyze the fact that you bought a Stephen Curry jersey on Amazon, record all of the games on DISH Network, and bought a plane ticket to the next game on Expedia.

The avatar would also know your food preferences, based on online searches. It could predict, with a high degree of accuracy, what you might say and do in many situations.

That might seem like a stretch, but it's possible your online footprint is already robust enough to create a rough facsimile of you. Your posts on Twitter, Google searches, photo archive, and even your text messages are like a blueprint. And, according to at least one expert, there are some signs that this information could be used to create an entity that lives on forever.

Patrick Kalaher is the VP of Tech Strategy at Frog, the design company. He told me the computational power required for an avatar is getting faster, cheaper, and more nuanced.

"To the extent that we can use a vast database of digital expression--what we have written, or otherwise expressed in social media through pictures, videos, and audio of us--to predict what we might say or do in the future, what we look like, and sound like, we'll be able to make that happen soon," he says. At the same time, the idea that this would be a perfect replicant (to use a word from Blade Runner) is a ways off.

Kalaher brought up the example of his own father, who died in 2001. There's enough data to build an avatar that matches his phrasing and opinions about basketball players, but it would be harder to predict what he might say about the recent presidential election.

One issue is that there are many complex systems in the human brain that we don't fully understand yet, especially those related to micro-biomes, genes and proteomes. Individually, we might be able to understand them, but their interactions with other chemicals has proved to be a challenge not only to analyze but to model effectively.

Another issue is that "who we are" is more than a question about our DNA structure. Today, the field of epigenetics (or gene expression) and proteomics (the study of proteins) adds more complexity to the equation. An avatar that replicated "who we are" would have to know about the extremely complex interactions between our genes and proteins.

"Consider that most of the DNA in and around our bodies doesn't even belong to us--it's in our microbiome, which is largely unknown at this point," says Kalaher. "All of these are several orders of magnitude more complex to understand than mere DNA analysis."

Kalaher says some form of an avatar, though, is entirely possible already. To create a rough facsimile, we won't have to model everything, only enough to seem convincing.

How would this help? I imagine a digital avatar working a bit like a Facebook page for someone who is deceased. You can go back and visit the page and see photos and posts. An avatar might be able to help loved ones to some extent, although it's obvious this bot would never serve as a replacement. Being human also involves touch, emotion, and feeling. Decades from now, an avatar that can express the thoughts and feelings of someone might at least bring back some memories and possibly help with decision-making (say, suggest the name for a new baby or where to live).

More than this, a digital avatar might help us cope with our own mortality. We would see that there is at least some chance our opinions and thoughts will live on. As we post on social media and write our opinions, we're building a future avatar, one post at a time.

Published on: Dec 20, 2016