Roger Schank Wants Your Child's Mind

For such a smart guy, Roger C. Schank could use a few lessons in tact. During a two-hour conversation in his sprawling, 41st-floor home on Chicago's Gold Coast, the artificial-intelligence guru turned educator brands most "edutainment" software as "crap," blasts student-testing methods as "measuring all the wrong things," and chides school administrators for "taking 30 years to realize that overhead projectors could move from the bowling alley to the classroom."

Given such repartee, it would be easy to dismiss Schank, director of Northwestern University's Institute for the Learning Sciences (ILS) and former chief of Yale University's Artificial Intelligence Project, as an academic gadfly. Mistake. Schank and his team of 150 scientists and graduate students at ILS are pursuing a serious mis- sion: using computers and multimedia software to revamp education and business training.

STORY THEORY. In Schank's view, America's schools badly need reform because they fail to make knowledge relevant to students. He blames large class sizes and curriculums that make students memorize the same facts in lockstep, which turns kids off learning and doesn't teach them how to apply knowledge productively. The best way to learn, he argues, is to get information just when you need it. And that means individual instruction.

Cheaper computers and information networks are the tools of Schank's would-be revolution. ILS, with an annual budget of $8.5 million primarily from the Pentagon's Advanced Research Projects Agency and Andersen Consulting, does cutting-edge research on how the human mind processes information, then applies the insights to develop multi-media software that can teach students individually. "Schank and the ILS are pioneers on a wild frontier" of education, says John R.B. Clement, an educational-computing expert at the National Science Foundation.

At the core of Schank's work are theories about story-based teaching and goal-based learning. Schank, like many educators, is convinced that stories are the most effective teaching tools. And based on his own theories of memory, Schank also believes that the most complete learning occurs when new information is presented just as a student needs to use it: That's when the brain indexes information most quickly.

To create an electronic, just-in-time teacher, Schank has developed software tools that interact with video. The ILS's ASK systems, as they are called, let students choose among predetermined questions about a topic or even type in their own queries. Special "case-retrieval" software gives students access to a video database of experts who relay "stories" about the subject. During such a session, indexing software continually changes the available answers and follow-up questions based on the previous question or response, or the student's shifting interests.

In this way, the ILS program Broadcast News is currently being used as a tool to teach political science at an Evanston (Ill.) high school. A team of students produces a TV news story on 1 of 10 constitutional issues, such as that raised by the Rodney King beating, which was tried as a federal civil rights case. As they write a script and edit film from a library of video clips and interviews with experts, students learn about political science. Use the word "alleged" when describing the accused police officers in the script, for instance, and the computer hits you with a text and video discussion of the presumption of innocence in American jurisprudence. Call the attack "racially motivated," and the software opens a dialogue about race relations in the U.S.

Once an initial program is developed, similar ones can be "built" upon it--at a much lower cost. ILS spent $120,000 to produce for Chicago's Museum of Science & Industry an exhibit that allows museum-goers to play the role of a genetic counselor to patients concerned about sickle-cell anemia. But future projects based on the same computer program as the Chicago exhibit will cost 90% less.

WEAK ON DETAILS. The same tools are valuable for employee training. "One of the problems that business has is that new employees come in with a school mind-set," says GE Capital Corp. Senior Vice-President James A. Colica. "They don't understand things, they just repeat them by rote." Based on Schank's work, he has created a multimedia simulation course to train new credit managers in loan underwriting. The software teaches "perspective and judgment calls, just like on-the-job training," Colica adds.

Schank's current fascination with learning is a far cry from the Brooklyn childhood where he spent more time throwing footballs--he would still like to play quarterback for the New York Giants--than he spent hitting the books. A bright but not particularly motivated kid, he graduated 322nd in his high school class of 670. Schank's early years as an undergraduate math major at Carnegie Institute of Technology weren't distinguished either. "I probably shouldn't have graduated," he recalls, "since I did precious little--other than play ball and play fraternity."

Luckily for the 48-year-old Schank, his growing interest in the relationship between mathematics and language provided the discipline to earn master's and doctoral degrees in linguistics at the University of Texas. From there it was a short jump into the nascent artificial-intelligence field, which was trying to develop computers that could simulate human reasoning and communication. In the late 1960s, Schank landed a computer-science teaching job at Stanford University. Then, as the director of Yale's AI Lab from 1974 to 1989, Schank hit the big time, attracting venture capital to start two AI software companies and stirring a heated debate over his views about how human memory is organized.

Colleagues concede that Schank's ideas can sometimes get lost in the furor caused by how he expresses his opinions. "Roger is a genius who rubs off on people--sometimes that's good, sometimes not," says David J. Urban, president of Compu-Teach Inc., the educational-software company based in Redmond, Wash., Schank founded in 1982. Agrees Stephen C. Mott, CEO of Cognitive Systems Inc., an AI company Schank founded: "Roger has a way of telling people they're full of it at the same time he's asking them to write a $1 million check."

Although he is brilliant at conceptualizing ideas, Schank can be weak on details, colleagues say. In 1987, while Schank was chairman, Cognitive Systems narrowly averted financial ruin after its then-chief financial officer made more than $1 million in unauthorized withdrawals and stock-index trades, according to Securities & Exchange Commission documents. Schank resigned from part-time management soon afterward. Concedes one admirer: "Roger is interested in the conceptualization and the plan, but once something gets under way, he isn't there anymore."

ILS, created five years ago, may be an exception. Andersen wanted to bolster Northwestern's computer-science department, so it offered $15 million to start the institute. Today, ILS employs an inter- disciplinary staff of teaching faculty from Northwestern plus graphic artists, video-production technicians, and graduate students. It houses 18 "fellows," who are employees of its corporate sponsors: Andersen, Ameritech, and British utility North West Water. After two years, fellows earn master's degrees and take Schank's thinking back to their companies.

Why the corporate tie-in? "That's where the money is," Schank says bluntly. Since learning concepts are the same whether the student is an adult or a child, companies provide invaluable testing labs--without the politics involved in using new educational materials in the schools. "All [that companies] are interested in is if it works and will it save money," says Schank.

SMARTEST MOVE. That's a major reason Andersen, which spends $200 million annually on training for its 30,000 employees, wants to extend its support of ILS for five more years. Schank's group developed a 36-hour multimedia PC version to replace Andersen's 80-hour business-practices course, previously taken by every new consultant at one of two training centers. In one segment, for instance, students play the role of a human resources manager for a manufacturing plant. They are faced with decisions about salaries, training, and promotion for a cast of employees whose performance histories can be investigated at the touch of a button. Anderson's top consultants provide plenty of advice in the form of video "war stories" culled from taped interviews. By no longer sending employees to its centers for training, Anderson saves up to $10 million a year, says Partner John D. Smith.

Corporate training may not be the only use for such tools. Baby Bell Ameritech Corp. has spent more than $1.5 million to sponsor ILS research in hopes that it will lead to interactive products for use on the Information Superhighway. "Roger is working on reshaping the way people interact with information, and that fits our vision of wanting to manage information for customers," says Ameritech Vice-President James A. Goetz.

Yet Schank's top goal remains better education for kids. He hopes to have ILS software ready for schools within three years, but some software developers doubt that Schank, a perfectionist, will produce software cheaply enough to make it commercially viable. Says one: "They've got decent products, but decent isn't what Schank is about. It may take someone else to bring these to market."

Whoever commercializes ILS's ideas, the technology is overdue. Artificial-intelligence experts "have promised things that probably can't be built, like the all-knowing machines that you see in Star Trek," Schank says. "But maybe we could build a machine that's good as a teacher." That would be his smartest move yet.

Before it's here, it's on the Bloomberg Terminal.