Email Print Share
Science Nation banner

November 9, 2009

Virtual Newscast: News at Seven


News Anchors: The Next Endangered Species?

Tired of the same old network newscast? Well, look out Katie, Charlie, and Brian! If researchers at the McCormick School of Engineering at Northwestern University in Evanston, Ill., have their way, newscasts could become a lot different. Scientists here at the Intelligent Information Laboratory are working on a project called News at Seven.

News at Seven is a computerized newscast. The story selection, the words, the graphic background--even the expressions of the animated newscasters, or avatars--are all determined by computer. With help from the National Science Foundation (NSF), the lab's co-director, Kristian Hammond, and his team are developing artificial intelligence that, in some ways, can think like a journalist.

"We can do it for everyone in the entire world and we can do it for organizations, and so an organization that can't afford a studio and performers and writers can actually build a News at Seven piece in a matter of moments and do it cheaply and, quite truthfully, do it easily," explains Hammond, who received additional NSF funds through the American Recovery and Reinvestment Act (ARRA) of 2009.

"ARRA funds have ended up being transformational to our work. We were initially able hire two full-time journalists to help expand the reach of our systems and a designer to help make the animations and interactions more fluid," he adds. "An upshot of this is that we're able to spin one of our technologies into a commercial entity that is now funded and is generating content into under-represented areas of sports and career advisory reporting."

Not just the news

Hammond's work is going far beyond virtual newscasts. He's also created a program that generates opinion. "It has a movie review dynamic and that dynamic comes from literally us watching shows like Siskel and Roeper where we saw two people interacting," says Hammond.

Computer science grad student Nathan Nichols generates the review by having the computer scan the Internet for images, clips and information about the movie "Watchmen." Seconds later, two virtual reviewers, one male and the other female, are going at it.

Zack, the animated male avatar who looks like a Generation X cartoon character, proclaims, "The music is perfect--especially the opening to Bob Dylan's 'The Times They Are A-Changin'." The female, who Hammond's team has named Niko, responds rather passionately, "No, they've made the most reverent adaptation of a graphic novel ever."

Hammond, revealing where his virtual reviewers get their comments and personalities, explains: "we might get a sentence from a blog, we might get a sentence from a review, and we might get a sentence from Twitter, and string them together."

The movie reviews generated here are already online at zap2it.com. But not all News at Seven projects are for public consumption, at least not yet. For example, there is the one Hammond is working on with grad student Patrick McNally--the two are actually attempting to program humor into the computer by developing the lab's own comic strip.

Understanding humans better

"People are almost annoyed and creeped out by the notion that we're trying to make a machine be funny, but we're doing it by understanding human beings better and how and why human beings are funny," explains Hammond. "Roughly half the time, comic material the computer spits out is incoherent anyway, so comedy writers should rest easy, for now anyway."

But, sports writers--that could be another story. By searching box scores from the latest Northwestern Wildcats baseball game, Hammond has the computer whip up an article that, when read aloud, could have been written by any competent sports writer. The byline simply reads, "By The Machine."

"I got to tell you, people who write sports stories are made anxious by it," says Hammond, who believes that the more computers are able to understand human behavior, the easier it will be for us to interact with them. "The very nature of our work is to understand people and to map that understanding onto the machine," he adds.

For people in the TV news business, the idea of having on air personalities who work 24-7 and never need new clothes, hair or make-up might sound appealing. Could tomorrow's star journalists come straight from cyberspace? We'll just have to stay tuned.

The research in this episode was funded by NSF through the American Recovery and Reinvestment Act of 2009.

Miles O'Brien, Science Nation Correspondent
Jon Baime, Science Nation Producer


Any opinions, findings, conclusions or recommendations presented in this material are only those of the presenter grantee/researcher, author, or agency employee; and do not necessarily reflect the views of the National Science Foundation.