Computers That Know What You Need, Before You Ask

Mar 17, 2014
Originally published on March 20, 2014 12:14 pm

We're already giving voice instructions to virtual personal assistants, like Apple's Siri. But artificial intelligence is getting even smarter. The next wave of behavior-changing computing is a technology called anticipatory computing — systems that learn to predict what you need, even before you ask.

Google Now, which is available on tablets and mobile devices, is an early form of this. You can ask it a question like, "Where is the White House?" and get a spoken-word answer. Then, Google Now recognizes any follow-up questions, like "How far is it from here?" as a human would — the system realizes you're still asking about the White House, even without you mentioning the search term again. It's an example of how anticipatory computing is moving the way we interact with devices from tapping or typing to predictive voice control.

"That's what is the next wave of computing, in my opinion," says venture capitalist Om Malik, who founded the technology news site, Gigaom.

Malik says smart virtual assistants using this technology are necessary to clear the digital clutter in our lives. The more we add apps and digital functions we need to perform on our devices, the more individually tapping or typing for each function becomes a hassle.

"As we become more digital, as we use more things in the digital realm, we just need time to manage all that. And it is not feasible with the current manual processes. So the machines will learn our behavior, how we do certain things, and start anticipating our needs," Malik says.

For example, if you're traveling, anticipatory devices fetch flight-delay or gate information automatically. Or they mesh the data in your calendar with maps to guide you from one appointment to the next.

During the NPR interview with Malik, we used an experimental app to help supply background information, photos and video on my tablet. The app — called MindMeld — listened in on our conversation and pulled up relevant information on my screen. For example, when I said Om Malik's name, his biographical information like a Wikipedia entry, news coverage of Malik and more showed up for me instantly. It's like Siri on steroids, capable of listening in on a conversation with as many as eight people at once.

"[Anticipatory computing] is gonna get good enough in certain areas that we're gonna wonder how we ever lived without it," says Tim Tuttle. His Expect Labs created the MindMeld technology.

"We now have all these devices with all these great sensors that can hear us, that can see us, that can know about places that we've been, places we're going, and those signals become the inputs that allow these intelligent assistants to find what we need without requiring us to type searches necessarily," Tuttle says.

Everything you say becomes a piece of data that anticipatory apps use to better understand you.

"The more they know about what you like, what you don't like, where you go, what you're talking, what you're reading, the better they can recommend something for you," Tuttle says.

It can seem kind of creepy — all our data going to these companies that control the technology. Malik says he's OK with trading in his data for the convenience promised by these computers. (He wrote about this trade-off in a piece for Fast Company.)

"Twenty years ago we were all squeamish about instant messaging. Then we got squeamish about Facebook," Malik says. "There is something inevitable about technology. It is scary, but at the same time, it is inevitable."

Tuttle says the change to these magic computers won't come that fast.

"There's still this mismatch of expectations that people have. They expect the Star Trek computer on day one," he says. We may not be quite there yet, but the era of magical computing is beginning.

Copyright 2014 NPR. To see more, visit http://www.npr.org/.

Transcript

AUDIE CORNISH, HOST:

Let's zoom in now on software that predicts what we want. That kind of artificial intelligence is all around us and it's quickly getting more advanced. Many of us give voice instructions to virtual personal assistants, like Apple's Siri. And as NPR's Elise Hu reports, some apps go further.

ELISE HU, BYLINE: When we throw out terms like artificial intelligence or robotics, it's easy to imagine WALL-E or R2-D2.

(SOUNDBITE OF ROBOT BEEPS)

HU: You know, something futuristic sounding. But artificial intelligence is already here in some interesting forms. Something called Google Now is one example.

OK, Google Now.

(SOUNDBITE OF A TONE)

HU: Where is the White House?

(SOUNDBITE OF A TONE)

UNIDENTIFIED VOICE: The White House is located in United States of America.

HU: What is the address?

(SOUNDBITE OF A TONE)

UNIDENTIFIED VOICE: The White House is at 1600 Pennsylvania Avenue, NW Washington.

HU: How far is it from here?

UNIDENTIFIED VOICE: To drive from your location to the White House is 2.2 miles.

HU: Notice I didn't type in anything with my hands. This was a voice interaction and when I asked what is the address or how far it was from here, Google knew I meant the White House. That's machine learning made possible by anticipatory computing, taking what you asked previously to know what you need next. It's also moving the way we interact with devices from typing or tapping to voice control and predict it.

OM MALIK: That's what is the next wave of computing in my opinion.

HU: Om Malik is a venture capitalist and founder of the tech news site Gigaom. He says smart virtual assistants using this technology are necessary to clear the digital clutter in our lives.

MALIK: As we become more digital, as we use more things in the digital realm, we just need time to manage all that and it is not feasible with the current manual processes. So the machines will learn our behavior, how we do certain things and start anticipating our needs.

HU: For example, if you're travelling, anticipatory devices fetch flight delays or gate information automatically or they pull up maps to guide you from one appointment to the next using the data in your calendar. As I'm speaking with Malik, an experimental app is anticipating the background information I might need for this interview. The app, called MindMeld, is listening in on our conversation and pulls up relevant information on my screen.

For example, when I said Om Malik's name, his biographical information showed up for me. Tim Tuttle created the MindMeld technology.

TIM TUTTLE: It's going to get good enough in certain areas that we're going to wonder how we ever lived without it.

HU: Tuttle has worked in artificial intelligence for years and he says predictive computing is about to take off.

TUTTLE: We now have all these devices with these great sensors that can hear us. They can see us. They can know about places that we've been, places we're going and those signals become the inputs that allows these intelligent systems to find what we need without requiring us to type searches necessarily.

HU: Everything you say becomes a piece of data that anticipatory apps use to better understand you.

TUTTLE: The more they know about what you like, what you don't like, where you go, what you're talk - what you're reading, the better they can recommend something to you.

HU: It can seem kind of creepy, all our data going to these companies that control the technology. Malik says he's OK with trading in his data for the convenience promised by these computers.

MALIK: Twenty years ago, we were all squeamish about instant messaging. Then, we got squeamish about Facebook. There is something inevitable about technology, you know. It is scary, but at the same time, it is inevitable.

HU: Tuttle says that change to these magic computers won't come that fast.

TUTTLE: There's still this mismatch of expectations that people have. They expect the "Star Trek" computer on day one.

HU: You remember that?

(SOUNDBITE FROM TV SERIES "STAR TREK: THE NEXT GENERATION")

MARINA SIRTIS: (as Troi) Tell me the location of Commander Data.

MAJEL BARRETT: (as Enterprise Computer) Lt. Commander Data now located in holodeck area 4-J. And as you can see, sir, it's pointing you that way.

HU: We're not quite there yet, but all knowing computers are pointing us that direction. Elise Hu, NPR News. Transcript provided by NPR, Copyright NPR.