Zite Geist: An iPad App that Learns
Artificial intelligence takes the leap from a UBC lab to Zite, a newly released iPad app.
UPDATE: On 30 August 2011, Vancouver tech startup Zite was purchased by CNN for an undisclosed sum (thought to be $25 million). The acquisition marks an aggressive move into digital delivery for the news giant.
It’s fair to say Zite surprised everybody. When the iPad newsreader launched on iTunes at the beginning of March, tech writers from such influential titles as the Wall Street Journal and Fast Company were amazed by the app’s easy-on-the-eyes layout and preternatural ability to learn users’ tastes.
Zite’s Vancouver-based team was overwhelmed by the 120,000 downloads the free app garnered in its first week. The company’s core group of five 20- and 30-something developers, mostly grads from UBC’s computer science program, stayed up for 38 hours straight in their Yaletown office to deal with the crush.
And, now famously, a large party of media heavyweights, including the Washington Post, Dow Jones, Time and the Associated Press, were shocked to see their content distributed on Zite for free, and without their hard-won online advertisements.
For Nando de Freitas, however, the most surprising thing about Zite isn’t the 12-page cease-and-desist letter these media monoliths sent, nor the resulting chatter over what it means for the future of the media business. It’s what isn’t being talked about that amuses the UBC professor credited with connecting the dots that led to Zite. “People are talking about this application and saying, ‘Oh it learned what I like,’ and no one is questioning the implications of this. There’s a machine that learns about you,” he says, with an incredulous laugh. “I was surprised it didn’t raise alarms. People just expect machines to learn, but actually getting a machine to learn is far from obvious.”
He would know. An associate professor of computing science and cognitive systems, de Freitas is an expert in machine learning and a big chunk of the brains behind Zite’s ability to “discover” users’ interests based on their search behaviour and social media patterns. It’s the secret ingredient that sets Zite apart from other news aggregators and allows it to crawl over half a million websites to find new sources its users will like.
De Freitas is also a matchmaker. Finishing a paper after hours in his office on campus, the soft-spoken 40-year-old remembers he was doing pretty much the same thing back in 2004 when Ali Davar, a gregarious law student at the time who is now Zite’s CEO, knocked on his door with an idea for an intelligent search engine. De Freitas put him in touch with Mike Klaas, one of his most promising PhD candidates, now the company’s CTO, and forgot about it.
That is, until it became evident the two were on to something with their first prototype, a search engine plug-in called Worio. Though featuring much of the same technology as Zite, Worio got lost in the vastness of the web, where newcomers to the browser game are bound to languish in Google’s shadow. It taught the team an important lesson, says de Freitas: “What matters is not the technology; what matters is how people react to it.”
In 2008 de Freitas took a sabbatical to work with his band of whiz kids and helped perfect the technology for the iPad. The platform offered them a chance to get noticed, and indeed they did. But it hasn’t necessarily translated into financial success. Like the media giants watching it closely, Zite has yet to figure out a business model complementary to its goal of disseminating content freely and easily on the web. The company is currently subsisting on $4 million in angel investments and government grants. “We’re hoping that one day we will become really important and have millions of users and partnerships with media companies, and then we’ll make money,” says de Freitas.
But for now de Freitas, a scientist at heart, is just pleased the world is seeing a practical application for technology that, in his opinion, should raise more questions about the impact of artificial intelligence than how the New York Times online will function behind a paywall.
“Machines that learn, appearing in our society – that to me is a big thing. And people take it for granted.”
Save over 50% off the newsstand price with a subscription to BCBusiness Magazine
