Impatient, noisy, and fiercely independent, computer visionary Roger Schank aims to build educational software that puts students in the drive's seat and fun back into learning
As perhaps befits a man who never received anything but D’s in school for classroom conduct, the oddly cantankerous yet charming Roger Schank, a world leader in artificial intelligence and (some think) visionary of educational software, insists that no one has the right to tell students what to learn or how to learn it.
Sitting in his corner office before a coffee table strewn with the Styrofoam refuse of a carryout lunch, the 48-year-old director of Northwestern University’s Institute for the Learning Sciences gleefully attacks state education mandates, standardized tests, and computer programs that ask students to “blast” verbs as if they were in some kind of electronic shooting gallery. The enemy, as Schank sees it, is the prescribed curriculum—a kind of speed trap designed to stop kids who are all too avidly chasing their own interests. The American way, Schank believes, is to let people travel in their chosen directions; learning, he argues, should be about the pursuit of happiness. “Fun is always valuable,” he says. “If it’s not fun, you won’t learn it.”
Creating computer programs that are both “fun” and educationally sound, as opposed to those that are only divertingly entertaining, has been the institute’s raison d’être since Schank, in conjunction with Northwestern and Andersen Consulting, founded it in 1989. Its educational software, still being refined, is currently in place in six Chicago-area school districts, and Schank and his colleagues are exploring ways to market it nationally. The institute also produces job-training software embodying similar educational principles for Andersen and other high-powered sponsors such as Ameritech and the U.S. Defense Department.
Students using institute-designed computer programs edit and produce newscasts, create dancing bears and flying fish, and cruise the interstate highway system. It’s all very different from traditional schooling, which, Schank says, is still about making people “remember stuff that won’t stick around.”
“Cultural-literacy people are the modern-day American Nazis,” he asserts. He places in this group all self-appointed standard-bearers of the American curriculum. “They’re sitting there saying, ‘I know what you need to know, and you’ll know what I say.’ Did you ever look at what’s on that list? I don’t know most of that stuff. It’s someone arbitrarily deciding that everyone should know this Eskimo folk tale. That’s ridiculous. What are they going to do? Drill it into peoples’ heads and test them to see if they know it?”
What then, if not a sanctified body of knowledge, should students study? Schank likes to talk about what he calls the “truck curriculum.” Most elementary school boys love trucks, he points out. Why not let these youngsters load them, drive them, smash them, and repair them? They’d have fun and learn something along the way. Smashing trucks and calculating load weights could, for example, teach students something about mathematics and physics. Transporting freight across the country could teach them something about geography, the nation’s highway system, and the duties of a dispatcher.
Of course, constructing such a learning situation in an ordinary classroom is problematic: Smashing trucks can be disruptive and destructive. Here is where the computer comes in. Schank believes that the computer’s interactive and simulative capabilities make almost anything possible. A kid can smash atoms or design a spaceship that will fly to Mars, just as pilots in a flight simulator—Schank’s favorite computerized invention—can combat wind shear without worrying about a fatal crash.
A computer “course,” then, according to Schank, can be created out of almost anything that interests students. Its goal should be to teach skills rather than to impart information, though students will of course acquire information as they need it to perform a given task. Schank contends—and cognitive research supports him on this point—that information is retained only if discovered in useful contexts.
In one of his many published articles (he is also the author of numerous books), Schank envisions a high school or college course called “Cure the Diabetic” that could augment or even replace the standardized lectures and textbooks of the traditional biology class. The computer program would let students consult with patients—the software having “stockpiled” a myriad of cases—perform simulated lab tests, and, after gathering and studying the results, suggest a course of treatment.
An institute software program paralleling “Cure the Diabetic” actually exists. That program, Sickle Cell Counselor, was created specifically for Chicago’s Museum of Science and Industry, which requested an interactive multimedia program to replace the frequently ignored wall-to-wall text. Sickle Cell Coun-selor asks the user to advise couples who are worried about passing the sickle-cell gene on to their children. After learning about a given couple’s concerns and medical histories, the user can ask a doctor, geneticist, or lab technician up to 40 questions about the disease. The user might inquire, for example, about the symptoms or who is at risk. He or she might ask a technician to demonstrate the use of a syringe and help “draw” blood. Once lab results are attained, the user, with the assistance of the software, calculates risk and then advises the clients.
According to research conducted by the institute, visitors spend an average of eight minutes at the exhibit; museum officials had hoped for two.
Schank insists that traditional schooling, in which the teacher makes students complete specified tasks, is a form of bullying that leads to abject conformity. This view has not always won him friends. During a recent trip to Asia, it won him downright disdain. Speaking to a delegation of Singaporean teachers, Schank told them, with his customary indelicacy, that their school system is “terrible.”
“A thousand people wanted to shoot me,” he says. “They think their kids are so well-behaved and informed. But when I asked them how many of them could still pass a basic biology test—the kind of stuff they demand their students learn—only three hands went up. They couldn’t remember any of it. `What,’ I asked them, ‘does that tell you about your educational system? If you passed all those exams but now can’t remember anything, then you know something’s wrong. We have to find a different measure of success.’ “
The thought of Schank being asked to address ultraconservative Singaporean educators is comically incongruous—a bit like the late radical Abbie Hoffman being feted at a Republican convention. Massive as an NFL nose guard, Schank is the Mr. America of the intellectual world: impatient, noisily confident, and fiercely independent—the living antonym of self-effacing. With a perfectly oval, bald head, Schank is literally an egghead but an egghead you wouldn’t want to mess with. Wearing a jet-black shirt buttoned to the collar, he’s Daddy Warbucks with a slightly ominous aspect.
Even as a child growing up in Brooklyn, Schank could not tolerate people telling him what to do. Preoccupied with one day playing quarterback for the New York Giants, he had little interest in classroom learning. Still, he thrived in elementary school, he says, because “I was always the smartest kid in class.” In high school, however, he foundered. Teachers actually expected him to do work—work he found meaningless—and he refused. “I quickly saw that the successful kids were the ones who were going to do everything just the way the teachers told them to do it,” he says. “I wasn’t good with authority figures and was going to do things my own way no matter what.”
After receiving his undergraduate degree at the Carnegie Institute of Tech-nology, where he claims to have majored in “fraternity,” he moved on to the University of Texas where he received a Ph.D. in linguistics in 1969. (“Only when you’re in graduate school,” he says, “can you study what you really want to study.”) Soon the burgeoning field of computer science captured his interest. He studied artificial intelligence at Stanford and eventually became a professor of computer science there. But Schank has always been much more than a computer scientist. At Yale, where he moved to in 1974, he was a professor of both computer science and psychology.
From the very beginning, he was concerned with the question of how people learn; this had to be understood, he believed, if computers were ever to become true educational tools and not just electronic play-boxes. While psychologists understood something about human cognition, they knew virtually nothing about computers. Computer scientists, on the other hand, were building increasingly sophisticated machines and programs but had little understanding of how people learn. At best, the resulting software was useful and scintillating but provided little information that couldn’t already be found in books. At their worst, the results were computerized workbooks—traditional schooling with glitz.
Schank’s goal, then, was to fashion a theory of learning that would drive the development of artificial intelligence. What he and his colleagues came up with sounds like vintage John Dewey: People learn best by doing, by drawing from both their experiences and those of others, and by failing.
Being allowed to fail is essential. Schools typically penalize failure, when, according to Schank, it is the one thing absolutely indispensable to learning. In the world of learning, there are no command performances, only a string of endless rehearsals; we make the inevitable mistakes, and then we correct them. As far as learning is concerned, we would do well to remain adventurous 2-year-olds, running around exploring this and that, getting into trouble, and then picking ourselves up to begin the process all over again. “Computers,” Schank writes in a recent issue of MultiMedia Magazine, “provide novices with a safe haven for making mistakes.”
In 1980, Schank, who was already director of Yale’s Cognitive Science Project, also became chairman of the university’s computer science department. By this time, he had long been creating educational software. But now, as his own children struggled with school, that quest consumed him.
“My kids were endlessly battling with the school system over the most silly things,” Schank says. “My kids would come back from school with these questions, and I’d say, ‘I don’t know why you should do it this way; why don’t you do it any which way you want to?’ Once, I was called into school because my son, then in the 4th grade, wouldn’t add with rods. You know—blue and yellow equals green. He already knew how to add and multiply, and he thought the rod thing was stupid. ‘Well, it is stupid, teacher,’ I said.
“But it wasn’t always the teacher’s fault; a lot of it was the dictated curriculum. One of my kids was forced to do syllabification. So I went to the teacher and said, ‘Why are you doing this?’ She said, ‘I don’t know. I tried doing something else, but they wouldn’t let me.’ “
This kind of frustration continued year after year. Schank’s son returned to his American high school after living in Paris for two years and could get only B’s in French, even though he was the only student in his class fluent in the language. His daughter had similar problems. One day, she came home complaining that whenever her teachers spotted an essay with her name on it, they automatically gave it a B. “I doubted that,” Schank says, “and so I suggested that she try writing the way the A kids did. She did that and sure enough got A’s. But she wasn’t happy. ‘I don’t like writing like that,’ she said. I told her she had a choice: She could get A’s or write as she wanted. ‘I’ll write the way I want,’ she said, and so, once again, she received B’s.”
“Now,” Schank laughs, “she’s a professional writer.”
Schank never intended to leave Yale, but Northwestern’s offer to create and head up the Institute for the Learning Sciences was impossible to turn down. Schank was told that he could handpick his staff, which would allow him to add faculty from a wide range of academic disciplines. Contracts to produce multimedia training software for Andersen and other corporations would supply the institute with steady funding. Best of all, he would essentially have carte blanche to design educational software that embodied the trial-and-error learning process. In 1989, he moved to Evanston, bringing with him many of his Yale colleagues. The institute now employs more than 150 professors, graduate students, re-searchers, and computer programmers.
But does Schank really believe he can build software that will change an educational system known for its intractability? “I don’t want to change it,” Schank says. “I want to replace it. But I know the political opposition is phenomenal, which is why we don’t run up the middle against the school system but rather around the end. We want to build computer-based courses the school system can’t possibly reject. A course in Japanese, for instance. There are no Japanese courses in this country. So, schools will say, `Gee, that’s a good alternative,’ whereas if we built one in French, we wouldn’t have a chance. Or how about astronomy or a senior-level course in physics? Seniors have already fulfilled graduation requirements, and schools don’t care what they do. Or how about a 3rd grade course in physics? There are no such classes, but there are a lot of kids who would love to smash trucks in computer simulations.”
When the institute designs a corporate-training program, there is little question of what the program should teach: skills that are needed in a profit-seeking organization. But the question is much hazier in the world of schooling. Schank himself has forcefully argued against an imposed curriculum. So how does he decide just what his computer programs should teach?
Out of necessity, Schank says, the marketplace has to determine the broad educational objectives of his software. “It’s political,” he says. “Someone provides the money, and we have to go along with that.” A grant from IBM, for example, led to the development of a geography program titled Road Trip and a biology program titled Creanimate, though Schank insists the institute had great latitude in determining the shape the programs would take. “But what would my answer to your question be if I could set the objectives?” he says. “It would be ‘Let kids go where they want to go.’ No one can write a list stating what they must learn. The name of the game is to find something that will interest them.”
Perhaps the most interesting of Schank’s ideas about the nature of learning and its relation to educational software is his all-abiding faith in the efficacy of storytelling. In his remarkable new book, Tell Me a Story (Macmillan), Schank, at different junctures, writes: “Hearing and telling stories is strongly related to the nature of intelligence”; “Performance in a conversation is an excellent measure of intelligence”; “The most you can expect from an intelligent being is a good story.”
Schank believes that what we know is, in essence, a compendium of all the stories we have ever absorbed, cleverly indexed so that we can recall them at the right times. If, for instance, we are thinking of quitting one job to accept another, we call to mind all the stories we have heard and read about people in similar situations. Based on our particular circumstances, we may hone in on a story about a man who had to decide whether to uproot his family or one about a woman who had to decide between a fulfilling job and one that is more lucrative but less emotionally rewarding.
But intelligence can’t be determined merely by the size of one’s collection of stories and the ability to be reminded of a particular one at a particular time. Intelligence, Schank makes clear, also involves the ability to evaluate and act upon the stories we know—that is, the ability to adapt existing stories to new situations. A school principal or company manager may know many stories involving sexual harassment, but the intelligent administrator, faced with a reputed incident in the workplace, will only take action when he or she has considered how certain recollected stories may or may not be relevant to the unfolding one.
What this means, Schank says, is that all educational software, whether for schools or for companies, should have strong storytelling capabilities because we learn best by considering vast numbers of stories in light of our own experiences. This storytelling dimension is most clearly evident in the training software that the institute produces for clients such as Andersen Consulting and the Defense Department. One program, titled “Yello,” turns students into sales agents for the Yellow Pages. In one scenario, the student-agents approach a roofing contractor and his wife who are about to pass their business on to their son. After hearing their stories, the students must decide on a selling strategy that takes into account the needs and problems the stories have revealed. If they wish, the students can call on veteran agents—all “in” the computer—to hear “war stories” about selling experiences that might help them on the sales call.
“In the end,” Schank writes in his book, “machines, like people, will have to be the repository of extraordinarily large numbers of stories in order to have something useful to say. Intelligence, for machines as well as for humans, is the telling of the right story at the right time in the right way.”
Schank freely acknowledges that it will be years before computer programs with such exceptional capabilities are available for use in education. Still, all of the software the institute is developing for schools has storytelling components.
In Road Trip, a sort of geography lesson on wheels, students use a mouse to take simulated car trips across the United States. Whether they travel aimlessly or purposefully, they must learn the meaning of certain cartographic symbols. When the students arrive at a particular destination, they can see stories in the form of video clips. In Harrisburg, Pa., for example, students may watch a scene from the movie Witness in which Amish men raise a barn. In New York, they may watch King Kong scale the Empire State Building or the 100th anniversary celebration of the Statue of Liberty. The video clips students choose to watch depend upon their interests. Current categories include sports highlights, movie clips, amusement parks, and historical footage.
The biology program Creanimate, designed to teach elementary students about animal morphology, works on a similar principle. The students can create their own animals, such as a fish with wings or a bee with a big nose. The students then view videos—capsulized “stories”—that illustrate the animals they have created and the purpose of the particular characteristics they have selected.
With Broadcast News, another institute program, storytelling is the very essence of the software. In one sequence, for example, high school students are asked to put together a news segment for March 3, 1991—the day of the Rodney King beating in Los Angeles. As assistant producers, they must construct an unbiased account—or story—of the event; the software program provides news text from The Associated Press and video clips from Cable News Network. Of course, what seems unbiased to one person may be blatantly prejudicial to another, as the students quickly discover, since the programmers have loaded the AP news account with questionable assumptions. The first sentence of the text reads, “A videotape of racist police officers beating a criminal has prompted the LAPD to investigate allegations of police brutality.” The programmers hope words like “racist” and “criminal” will raise red flags and encourage students to seek expert counsel. They might turn to Northwestern professor of journalism (and former NBC president) Bob Mulholland, who would ex-plain that this use of the word “racist” would not meet his criteria for fairness. Or they might call on a political scientist who would suggest that this use of “criminal” could be considered both inflammatory and unjust.
As he demonstrates Broadcast News, institute programmer Frank Luksa says that the point isn’t so much to teach students about journalism—though that is a welcome consequence—but rather to have them explore “the issues behind the news.” In deciding how much weight to give a particular element of the story, the students need some understanding of criminal law, race relations in Los Angeles and the United States, and the protocol police are expected to follow in such situations.
While the program provides some source and background material, the software, Luksa makes clear, is not designed to render the teacher superfluous. To the contrary, the teacher, depending on the particular goals of the course, might initiate a discussion on the Bill of Rights or the moral responsibilities of the mass media. Teachers who want to discuss the possible socioeconomic roots of the ensuing riots might even have their students undertake a demographic analysis of South Central Los Angeles. In short, there is no reason why an exploration of the subject should begin and end with the computer program.
“We know that kids like the program,” says Luksa, noting that it is being tested with 11th graders at a local high school. “One group was so immersed in putting together a broadcast that they actually blocked the door so that teachers couldn’t interrupt them in their work. They have a lot of fun with this.”
Staff members at the institute seem to have internalized Schank’s almost epigrammatic “Whatever’s fun is valuable.” While it’s easy to share their enthusiasm for the joy of learning, however saccharine such talk sometimes sounds, one has to wonder just how long a program like Broadcast News will remain “fun” for students. Will it have educational staying power, or will it, after a day or two, sit unused in a classroom corner?
As if anticipating the question, the software creators quickly point out that the program is an “advanced prototype” with clear limitations. By relating the news events of only a single day, the software simply does not tell students enough stories. They have no inkling, for example, of the many other stories that unfolded in subsequent days, months, and years, such as the forced resignation of LA police chief Daryl Gates and the acquittal of the accused officers. Furthermore, the almost exclusive reliance on experts from the “educated classes”—journalists, political scientists, lawyers, and the like—conveys a sort of academic arrogance. Without the stories of South Central residents—a black minister or a laid-off laborer, say—the Rodney King episode loses its emotional resonance, threatening to become an entertaining but all-too-intellectualized exercise.
Incorporating more stories into the software will address some of these concerns. So will an “authoring” system, currently being developed, that will enable teachers with limited computer skills to create and enter Broadcast News segments into the system.
At a six-week institute workshop this past summer, eight Wilmette (Ill.) teachers and one school administrator used such an authoring system to develop interactive programs—modeled on Sickle Cell Counselor—for elementary school students. In one program, students consult with arson investigators and detectives to get at the cause of a suspicious fire. In another, students advise a disabled student who is trying to get into a baseball game how to negotiate the crowds. In yet another, students counsel a child dealing with the travails of friendship. These programs are now being tested in a handful of schools.
Generating their own concepts, scripts, and videotapes, and then “transferring” them into the computer, was not an easy task. In fact, several of the participating educators say they have never worked so hard in their lives. But they had the feeling they were pioneering a potent educational tool. Jay Fry, principal of Highcrest Middle School, says the only real limitations of the technology that he can foresee are those of the people putting the programs together. And he believes those people ultimately will be students as well as teachers. “Creating goal-based scenarios”—the institute’s term for interactive programs requiring the sort of trial-and-error problem solving Schank so advocates—”must become part of the goal-based scenario,” Fry says. “Imagine, for instance, high school kids creating computer programs for middle school students.”
As enthusiastic as Fry is about the work he and his fellow educators performed at the institute, he sees computers as a powerful reinforcement of good teaching, not as the locus of any given course. Asked if the institute’s educational software was doing the sort of things with students that teachers should be doing anyhow, Fry says, “Absolutely.”
“Computers are a valuable tool,” he explains, “but they’re not the answer to the so-called educational crisis.”
Hanging over any discussion about educational technology is a question of considerable consequence: Is technology an important classroom tool or should it become the very focus of schooling? It’s a question many educators ponder but rarely ask, fearing, perhaps, that there will be little room for teachers in the age of computers.
Their uneasiness is understandable. Over the past few years, a number of education critics have argued that technology should and will play an increasingly large role in schools. At the extreme—some might say the fringe—is Hudson Institute scholar Lewis Perelman, who in his 1992 book, School’s Out, portrays schooling as an expensive superfluity. In the world he envisions, computers, like pet dogs, are everywhere. Whether at home, on the beach, or in the stands at a baseball game, kids are brandishing their computers, scanning Civil War battlefields and calculating the trajectory of a satellite on its way to Mars.
It is quintessentially American to hanker after the new and improved, and nothing seems more new and improved than the almost incalculable manifestations of computer technology. Compared with it, the teacher can seem a dusty and archaic figure. Yet there is no reason to consider Schank’s educational software, or any other software, as a threat to teachers—good ones, anyhow.
This is not to imply that software programs like those developed by Schank and his institute are not entertaining and instructional. What parent would be content with their children merely memorizing state capitals and coloring in maps when they could travel and navigate the country using Road Trip? And who would have their children only memorize animal group characteristics when they could use Creanimate to discover for themselves how even the most unlikely traits can ensure a species’ survival?
On the other hand, who, with the possible exception of Lewis Perelman, would want their children to have an education consisting only of Road Trip, Creanimate, Broadcast News, or any other combination of computer programs? Schank is perhaps being a bit disingenuous when he says, “I want Broadcast News to be the whole course.” Would he, so critical of his own children’s education, really want his kids to take a course defined solely by a computer program? After all, Broadcast News, as impressive as it is, cannot allow “everyone to go in different directions,” Schank’s stated ideal. It provides a script, one that permits students to make a wide variety of decisions, but a script nevertheless. They can edit the existing text but cannot, within the confines of the program, create their own.
Furthermore, there is no assurance that students will use technology in ways that are educationally meaningful. Institute staff members experienced this firsthand while testing Creanimate, which graduate student John Cleave describes as usable but still in the experimental stages. “We discovered in our testing that too many kids were watching the videos,” Cleave says. “The hope was that the videos would encourage kids to work out issues pertaining to survivability, but that’s not always the case. Some ‘video-hoppers’ want to do nothing but see videos, trying to turn the software into the Discovery Channel. We have to change a few things so that kids are coerced to apply knowledge and not just watch videos.”
Of course, if technology has limitations, so do teachers. In a given subject area, a teacher can store and dispense a minuscule amount of information compared with a computer. But computers are tools, and good teachers, of course, are not. They are far too nuanced for that role, and it is their job to respond to the nuances—the immeasurable quirks, if you will—of the students they teach. Like Schank’s ideal computer program, good teachers are repositories of many stories that collide and mesh with the experiences of their students. Good teachers know, in a way a machine never can, how to ask the right questions and tell the right stories at the right times. They are in a position to know what makes those video-hopping students tick. They know, or should know, something about the backgrounds of students who insist on retaining the phrase “racist police officers” in their Broadcast News scripts, regardless of how often they hear the expert’s sound bites.
Schank doesn’t think for a minute that machines will usurp the roles of teachers and schools. “I don’t believe that anyone should be sitting before a computer all day,” he says. “Lewis [Perelman] seems to forget that children are social people and that the school serves a useful social purpose. And there is a need for authority figures—teachers modeling right and wrong.
“But nurturance is the real key in teaching children. You need someone who genuinely enjoys playing with the software and wants to discuss it with the kids. What you do not need is an expert. The idea that teachers have to be experts is from 1910. Today, the experts are all in the box. No one could do all the physics assembled in the machine. So, let the teacher learn alongside the children.”
In suggesting that the computer be the expert and the teacher the nurturer, Schank is making the technophile’s distinction: The computer does the thinking, the teacher the caring. Although Schank clearly intends no offense, it’s hard not to think of the shift from teacher-as-expert to teacher-as-nurturer as anything but a demotion. A nurturer without some expertise is a care-giver, a custodian of the emotions. Such a person cannot be an authority—not when “all the experts are in the box.” While we want teachers to be nurturers, we also want them to be much more: We want a writing teacher to teach writing, a history teacher to teach history, a French teacher to teach French. We do not want them merely to sit with our kids, learning by their sides.
But what can a teacher teach kids that a computer can’t? Probably nothing if the teacher is a bad one and a lot if the teacher is good. The bad teacher is a textbook always going out of print; the good teacher is a series of nuanced responses. The bad teacher knows facts but not stories; the good teacher asks the right questions and tells the right stories at the right times. The computer might well replace the former, but it will only be a tool of the latter.
Vol. 06, Issue 04, Pages 22-27