Imagine that you're driving along an unfamiliar road, alone and exasperated. "Where the fuck am I?" you say.
Your car answers in a soothing voice, "Five and a half miles south of Gurnee. Looking for Six Flags? They have water rides, and they've just renovated Splashwater Fa--"
"No--I'm fucking starving."
"Would you like directions to Long John Silver's?"
"Fuck yeah I would!"
"All right, potty mouth, here's what you do."
You're driving, say, a 2009 Taurus. It knows, or seems to know, that you lose your way easily, that you swear like an Osbourne, and that you enjoy water parks and heavily fried seafood.
Hollywood has been promising talking vehicles--and sassy robotic maids and irrepressible mechanical pets--for quite some time, and it may be Chicago's Jellyvision that will first deliver them. The company has already partnered with conglomerates Ford and Philips Electronics to create interactive advertisements, and the Jellyvision Web site offers slick demos of other possibilities: a sarcastic virtual tour guide for the business traveler lost in Chicago, a spookily real simulated phone operator assisting a catalog shopper, a straight-faced investment planner coaxing financial goals from a reluctant investor. In each case, the dialogue involves a person interacting with a series of prerecorded phrases, but the discourse feels and sounds completely natural.
Jellyvision doesn't call it artificial intelligence, but the demos do evoke sci-fi novellas with machines-take-over-the-world endings. Jellyvision insists they're just harmless snapshots of a better future.
"We never wanted to be a gaming company," says company president Amanda Lannert. Yet Jellyvision--which began its corporate life in 1989 as a company named Learn Television, producing educational films such as The Mind's Treasure Chest--was a crazy success as a gaming company. Only three years ago it employed 70 people and was producing best-sellers such as You Don't Know Jack and Who Wants to Be a Millionaire? You Don't Know Jack, which debuted in 1995, used a smart-ass game-show host to propel players through a trivia game that's like Jeopardy on amphetamines, more Norman Lear than King Lear. Who Wants to Be a Millionaire?, based on the TV show, became the fastest-selling CD-ROM in history. But Jellyvision saw investment veering toward gaming systems such as Xbox and GameCube. "These games," says Lannert, "sell platforms and guns and girls."
And Jellyvision isn't in the guns and girls business. "I truly believe Harry Gottlieb is a visionary," says Lannert, referring to the company's 37-year-old founder. His version of man-machine interaction isn't achieved through complicated circuitry or runic mathematical formulas, but through the careful, intelligent work of writers and performers embedded in machines, which interact with the rest of us in preplanned yet seemingly spontaneous ways.
When the principles of Interactive Conversation Interface--shortened to iCi, or "icky" in company parlance--had organized themselves in Gottlieb's head, he spread the gospel to his company's creative staff and laid off most of the other workers. "We gave people nine months' notice," Lannert says, "nice parting gifts." The remaining 15 employees, occupying a newly spacious office near Goose Island, turned to changing the way we interact with the world.
"It's so not HAL," Gottlieb says, referring to the computer in 2001: A Space Odyssey. "I mean, the idea of human beings talking to machines and machines talking to human beings is an old idea. It's been around science fiction for a long time. It was begun to be thought of in a much more practical way in the 50s--these guys in AI who were like, 'Yeah, we're gonna figure out how to make these computers be more like human beings.' Like there's some way of getting there. And that approach, which is a computer-science approach, is really what's dominated the efforts to make machines more humanlike.
"Our approach is an audiovisual approach, a television and film approach, which is much more coming out of not writing complex algorithms, but coming out of having writers writing dialogue and actors performing the dialogue and creating a very complex branching structure where every point along the way has been crafted by an artist."
As Gottlieb talks, his hands are in the air, manipulating imagined blocks of recorded narration. "You put this little piece of audio here and put this little piece of audio next to it," he says. "Then if they say this here, then I would logically say this here. And then if they'd said this other thing before this, then probably I'd say this. It's called concatenation. And y'know, there's just a way of doing it smoothly. With You Don't Know Jack, which is a simple example, it was relatively repetitive. You have the same basic sequence over and over again. 'The next category is Michael Dukakis and baked beans. It's going to be worth $4,000. Put your fingers on your buzzers. Here's the question.' That's five different audio files that may have been recorded months apart. It's a new kind of performance. It's different performing onstage than it is performing for television, different performing for film, and it's different performing for radio, and it's different performing for this."
Asked about the origins of iCi, Gottlieb says he doesn't recall a singular revelatory instant. "In 1993, '94 I started doing my own little R & D exploration," he says. "'R & D' makes it sound like it's so much more official than it was. I mean, I was sitting in my bed in my underwear with my PowerBook--so R & D--and trying to get at this idea. My background's in film so I'm interested in communication. Video games? Not really communication but very interactive. When you think about a television show there are so many different genres, there are so many ways to communicate through television. It's way open-ended. It's agnostic to content, it's agnostic to style. Same with radio, same with newspapers. So these games didn't feel like they were describing a generalized approach to how you communicate in an interactive environment, and the adult-oriented ones were essentially fancy ways of navigating to find sound and text and video. And that's basically what the Web is. I mean, the Web is like this giant multiuser database on some level.
"So the question is, 'What is something that's interactive, that's really about communication, that is fundamentally audiovisual the way television is fundamentally audiovisual?' And that thing is conversation. What is 'interactive'? Using words, it's audiovisual, it's a conversation that can have visual elements to it if need be, if you're not over the telephone. And given that, how does one go about creating the feeling that you're having a conversation with a machine?"
Gottlieb seems genuinely baffled that he and his company are out front in developing this kind of man-machine interaction. "You look at the demos and it's like, doesn't it seem so obvious?" he asks, not waiting for an answer. "Producing them is hard, doing them well. But the end result for the user is so simple, which is the same as You Don't Know Jack. That's the way it should be. It should be hard for the people who are producing it and really easy for the people who are experiencing it."
Gottlieb has written a 79-page iCi manifesto, "The Jack Principles," which he says will soon be published--though he won't say where or by whom. "When you read 'The Jack Principles' after looking at the demos," he says, "you're going to be like, 'Jesus, this is the most obvious thing in the world.'"
"The Jack Principles" reads like a primer on human conversation written for a species that's slightly smarter and seeking to take us over. The headings in the table of contents include "Maintaining Pacing: Limit the Number of Choices the User Has at Any One Moment," "Creating the Illusion of Awareness: The Actual Time and Space That the User Is In," and "Maintaining the Illusion of Awareness: Use Dialogue That Conveys a Sense of Intimacy."
Jellyvision's creations are eerily close to the real thing. Gottlieb isn't hoping to create artificial intelligence, but to simulate human intelligence in a relaxed way that replaces text-based communication wherever and whenever suitable. And he clearly believes his company's creations can trump GUI, or graphic user interface, the point-click-drag function that still dominates relations between humans and computers.
"What is audiovisual about the Web is they'll put pictures on or a little animation, so that these are essentially illustrations to text," he says. "It's good for filling out forms, ordering things. But if I'm trying to sell you a car, or if I'm trying to give you investment advice, or if I'm trying to help you pick out a college--the kinds of things that a really well-informed human being sitting with you would do much better than a brochure on the screen--then there's an opportunity to meaningfully use our approach to interactivity. It's never going to be as good as the best possible human being--who also has audiovisual things at their fingertips so they can illustrate things--but it's a way of sort of bottling that, within some parameters, to make that available to millions of people one-on-one simultaneously."
Gottlieb thinks Jellyvision's interactive designs might play a role in society soon. "We hope that what we're doing is pioneering a new form of communication that is as fundamental as newspapers, magazines, radio, television, film, Web sites," he says. "We think that interactive conversation--or whatever it ends up being called--fits in that pantheon. And how many more of those are there?"
Art accompanying story in printed newspaper (not available in this archive): photos/Cynthia Howe.