Would you give your wireless carrier permission to listen in on your phone calls? Telefónica, one of the world’s largest mobile carriers, is testing a technology that can understand conversations and quickly pull up relevant information. If that info turns out to be useful, customers may want to invite it to listen in.
The technology is being built by a small San Francisco startup, Expect Labs, which is announcing strategic investments from the venture capital arms of three weighty backers today: Samsung, Intel, and Telefónica Digital, the business unit the telecom company launched in 2011 to unearth new revenue opportunities. The size of the investments was not disclosed.
Expect Labs has attracted attention because its technology is in line with the general direction that search technology has been taking with the advent of wearable computers such as watches and glasses, and Internet-connected cars and TVs. Rather than wait for users to search for something, the new technology offers up info that it thinks the user might need. The Google Now software for mobile devices, for example, already monitors its users’ locations, search history, and e-mail to call up traffic reports and other information. But Google Now doesn’t mine old-fashioned voice conversations yet. In October, Google Ventures invested in Expect Labs as part of a $2.4 million financing round by the startup.
Expect Labs has spent more than two years developing artificial intelligence technology that can parse the meaning of real-time conversations (Apple’s Siri, in contrast, can interpret only relatively simple spoken commands). Expect Labs’s “anticipatory computing engine” extracts the most relevant terms and uses them to offer potentially helpful information (see “Smart Assistant Listens to You Talk, Fetches Info Automatically”). For example, if two friends are having a discussion about grabbing some Thai food, it might call up reviews of nearby restaurants. If a company’s revenue comes up in a videoconference, it could display recent revenue charts. Expect Labs has built an app, called MindMeld, to demonstrate to partners how this works.
“The great thing about the platform,” says Telefónica Digital vice president of investments Christian Craggs, “is that voice is translated into text, and that is translated into an element of data.”
Telefónica, Craggs says, intends to test the technology internally and with some of its customers over the next six months. It could introduce it into products as soon as next year, he says. The technology could also appear inside TuGo, the company’s voice app for subscribers, or as a tool for customer service agents or doctors. Telefónica—which operates mainly in Europe and South America—will also explore how it could use the technology to create targeted advertising, he says. In all cases, he emphasizes, customers would have to explicitly opt in. Their conversations would be analyzed in real time, but not stored on company servers.
Expect Labs’s mix of strategic investors also illustrates how previously distinct business models are shifting in the wireless world.
For example, Expect Labs CEO Timothy Tuttle says that electronics maker Samsung is interested in his company because, in the smartphone and smart TV market, owning and controlling software and services is becoming more important—as Apple has shown (see “Why Samsung’s Man in Silicon Valley Uses Apple Devices”). Similarly, with traditional voice service becoming commoditized by the Internet (see “Do You Really Need a Voice Plan with that Fancy Smartphone?”), Telefónica Digital is looking to mine new revenue from the company’s existing assets. This includes using its wealth of data and access to customers to launch a market research service (see “How Wireless Carriers are Monetizing Your Movements”).
“Google would love to have the amount of voice information that these telecommunications companies have every day,” says Tuttle.
All that data will also help Expect Labs improve its technology, which is still a work in progress. By tapping into more voice conversations in its work with Telefónica, its algorithms will have more testing data to improve their predictive abilities.
Still, Tuttle says, the technology will take a few years to become mainstream. “We’re a small company that’s trying to solve a very hard problem,” he says.
How a Russian cyberwar in Ukraine could ripple out globally
Soldiers and tanks may care about national borders. Cyber doesn't.
Meet Altos Labs, Silicon Valley’s latest wild bet on living forever
Funders of a deep-pocketed new "rejuvenation" startup are said to include Jeff Bezos and Yuri Milner.
A horrifying new AI app swaps women into porn videos with a click
Deepfake researchers have long feared the day this would arrive.
Meta’s new learning algorithm can teach AI to multi-task
The single technique for teaching neural networks multiple skills is a step towards general-purpose AI.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.