The Future of Learning: An Interview with Alfred Bork
Educom Review: What changes do you envision for education and learning in the future?
Alfred Bork: I prefer to talk about learning rather than education or teaching. Only learning counts in the long run. Learning focuses on the student, not the teacher.
The current systems of learning are weak at all levels, even in the developed countries. In much of the world, little learning exists for most people. This situation is intolerable. Everyone should learn to their full capability.
Our views of learning come from a time when the world population was much less than it is now. The question today is, How are we to help six billion people learn, from birth to death? We require new models. It seems to me unlikely that schools and universities, as they exist today, will survive. We must be bold in addressing these problems.
Almost all learning is now at a fixed pace; a "course" is the same length for all students. I believe it was Patricia Cross who said that our present system keeps the time the same and varies the amount of learning but that it should be the other way around. Students should work at a given topic until attaining mastery. Students, with different interests, backgrounds, and learning styles, learn at different rates. Current educational structures make individualized pacing difficult, suggesting that these structures must change drastically.
Another major difficulty today is lack of equity. Some people have very good learning opportunities, but many, particularly the poor, do not. We need more equitable opportunities.
ER: What else is wrong with current models of learning?
Bork: The current dominant models of learning assume that the task is that of supplying information to students. So "learning" as it is usually found today -- in lectures, video, and web experiences -- is viewed as transmitting information from one person to another. Mostly we test on memory of this information, often with horrible methods such as multiple choice. We ignore the students who do not learn or who learn only partially with the information-transfer model, assuming that they are not bright or are not working hard enough. We neglect important high-level skills such as problem solving. These existing models are inadequate for the future.
New approaches to learning can resemble the interaction between a student, or a small group of students, and a skilled tutor. Such learning is highly active for both the student and the tutor. It does not offer the same approach to each student, as a lecture does, but is individualized to the needs of each student. Future learning should look continually for individual learning problems and offer help for these problems. Students should have adequate time to learn the material well.
Our delivery systems too must change. To reach very large numbers of students, essential for the future, we need effective distance learning. Students should be able to learn anything, at any place, and at any time. Current models of distance learning, particularly in the United States, are weak. They are often based on the information-transfer model, not the highly interactive model just suggested.
ER: How can computers be integrated into learning models?
Bork: Given the large and growing number of people, and the need for lifelong education, the best hope we have for individualizing learning is with the use of the computer, with adequate learning programs. The situation is similar to that in a popular television program of the 1970s, "The Six Million Dollar Man." In the opening sequence Steve Austin, a test pilot, crashes. The doctors viewing his battered body say: "We can rebuild him. We have the technology." Similarly, we can rebuild learning with the computer. But this involves the development, evaluation, improvement, and distribution of very large amounts of learning material -- activities that are almost nonexistent today. We do not even have trial projects investigating the type of learning I believe to be essential for the future.
I began using computers in learning situations in about 1958 with an IBM 1620, so there is a long history of computers to aid learning, but not continuous progress. I see most current uses of computers in learning as inadequate, often driven by technology issues rather than learning issues. The main advantage of the computer -- the possibility of wonderful human-computer interaction to assist learning -- is seldom realized.
ER: You mentioned a "highly interactive" learning model. What do you mean by "highly interactive"?
Bork: We can find the answer to this question by examining the student-tutor model already mentioned. Interaction is not a yes-no factor, but we can have varying degrees of interaction.
There are at least three factors in highly interactive learning, both with and without computers. The first is that interaction should be frequent, as in human conversation. Our experiments with computers in public libraries suggest that the student should do something meaningful, such as answering a question, at intervals of no longer than twenty seconds -- very different from the passive lecture environment. Such an active environment maintains student interest for a long period of time, even with difficult learning material.
Even more important is the quality of each interaction. This is more difficult to measure but is often clear in practice. We need highly interactive units, approaching human interaction, if we are to look continually for student learning problems. High-quality human interaction uses a critical human invention, one unique to human life -- our languages. Thus interaction in this country should be in English, in a two-way dialogue. Other countries will need other languages.
Several forms of high-quality interaction are possible. Perhaps the most important for computers in learning is the Socratic format, with the computer asking questions and the student replying in unrestricted English. In the future, students are likely to respond by speaking to the computer, a format that is already practical. Interaction by pointing or using multiple choice is of very poor quality; humans seldom communicate entirely by pointing!
The third important factor in interaction is long-range memory. A human tutor remembers the learning styles and past problems of students. We can provide a similar facility with the computer, keeping detailed records on student performance and using these records in making decisions about what is next to be presented to the student. Eventually these records will need to be world-accessible as people move around the world.
ER: Why do we need computers in highly interactive learning models?
Bork: In our traditional learning environments, some students learn and some do not. It is this second group of students that we want to help. Everyone should learn! This should be a critical factor for the future of learning, essential for the survival of humans.
A tutorial approach to learning makes it possible for everyone to learn. But before computers, this approach was too expensive for all but the very wealthy. Even if it were economically possible, we could not find enough excellent human tutors for all students.
The key to success is that we must determine, on a moment-to-moment basis, just what the student knows and just what learning problems are occurring, so that explicitly tailored help can be made available. This must be done very frequently, not just a few times during the student's life in learning in a particular area. Experience indicates that we can do this with highly interactive computer materials. We need more experimental work.
ER: Will highly interactive software require the use of artificial intelligence?
Bork: NO! We have been developing such material for thirty years at the University of California, Irvine, beginning with much weaker computers than are available today, without using approaches from artificial intelligence. This is not to say that artificial intelligence will not eventually be important for learning, but we can do without it at present. The key to developing effective computer-based learning is not the integration of particular computer tactics, such as artificial intelligence, but the use of skilled teachers, working in groups, in the design process.
ER: Can you describe some examples of highly interactive software?
Bork: About fifteen years ago, our project at the University of California, Irvine, developed a set of ten programs that IBM marketed as the Scientific Reasoning Series. These programs are still more interactive than almost all learning software developed today. But the graphical capabilities, done so long ago, now seem primitive.
One of these programs is Heat. The aim is to allow students to develop the important concept of heat, as distinguished from the often confused concept of temperature. It uses the student's own everyday experience. It begins immediately with a question, with no preceding text. The introductory question is, "How do you measure your own body temperature?" Then the computer waits for the student to type an answer, prompting if nothing happens after a few seconds. The entire program proceeds in this conversational fashion, with computer questions and student answers.
A second program in this series, Families, aims to let students discover the Mendelian laws of genetics -- not to be told the laws, as in most existing learning models. A mating simulation for the "Nors," the imaginary creatures introduced in the program, is central. But this simulation is less than 5 percent of the code, since students will usually need much guidance in using the simulation to find the laws. We want everyone to experience the joy of discovery.
ER: Should computer-based materials be added to existing lecture-based courses?
Bork: No. The problem of almost all modern learning is the lecture, a noninteractive way of learning. Even an excellent "sage on the stage" is not effective for all learners. University grades show that many students do not learn from lectures.
A lecture-driven approach could be a little less bad with added computer material, but not much. And the computer would be an add-on cost, not leading to lower costs for learning. To develop highly interactive courses, we should not base them on lecture courses.
ER: Are all courses better taught with interactive computer materials?
Bork: This is an empirical issue, only decidable with experimentation. But I believe any course currently taught in sections of more than about fifteen students can be learned far better in a highly interactive computer-based course,assuming well-developed and well-evaluated learning material. The critical factor is that we can react to individual student problems, even with large numbers of students. The computer probably will not compete well with very small groups taught by excellent teachers who follow an interactive approach. But these groups are rare.
ER: How should courses using highly interactive software be structured?
Bork: New ways of learning allow new learning structures. I do not say "courses" because in a full self-paced distance environment, learning will be a continuous process from birth to death.
The key concept for structuring highly interactive learning experiences is the Benjamin Bloom concept of mastery learning. Since the goal is for everyone to learn everything to the mastery level, grades will no longer be useful. We check frequently for mastery of the learning material, offering additional learning sequences if necessary and checking again. A student who has not learned in one way probably needs a different approach, rather than another go-round with the material that was not previously successful in assisting learning.
In such an environment, learning and evaluation are no longer separate activities but are part of the same process, intimately blended. So the student is not conscious of taking tests, and we avoid the problems of cheating.
Learning in a highly interactive environment can also be constructivist. I mean "constructivist" as guided discovery, with students discovering their own knowledge. Thus a physics student discovers the Newtonian laws of motion rather than being told them.
Such learning must hold the student's interest. We need to determine, through evaluation, that the student stays at the material. This does not imply that entertainment is necessary; motivation should be intrinsic to the learning material. As I already mentioned, highly interactive learning is intrinsically motivating. Motivation is particularly important in a distance-learning environment, since none of the "threats" of the classroom, such as low grades, are available.
A mastery-based computer segment could also offer human contact. Small groups could work together, either locally or remotely via electronic communication. Parents might participate with students. Other human help might also be provided.
ER: What are the steps involved in producing such learning structures?
Bork: When we first created highly interactive software over thirty years ago at Irvine, we looked at existing authoring systems. Since they were, and still are, mostly directed toward supplying information, these were inadequate for creating highly interactive software.
At that time we were working mainly on physics material. The first interactive program we wrote allowed students to derive the law of conservation of energy, for mechanical systems, beginning with the Newtonian laws of motion. In almost all existing courses the teacher does this derivation on the blackboard, or the students read the proof in a book. We wanted, and still want, students to discover their own knowledge through guided discovery, an effective way to learn.
We had to develop our own system for developing such learning materials. The design process we developed for that program is still the basis for our system. We were joined by computer science colleagues at the University of Geneva, in Switzerland. Software extensively supports the system.
We follow four steps to create highly interactive software. The first is project management. The second and most important step is design. We want a full specification of how the learning sequence is to behave with students. Our designers are excellent teachers in the content area involved, teachers who frequently work directly with students. The first phase in step two is overall design, planning the full learning sequence. Detailed design follows, conducted by groups of four or so teachers. Each group typically works together for about a week. An extensive project needs many such groups. We seek the best teachers we can find, often not in Irvine. The design groups decide what students are to see and hear, including media, what questions are to be asked, how free-form student inputs are to be analyzed, and what choices about future learning sequences are appropriate for each student. They also decide what information about students is saved and how saved information is used in the student-computer dialogue. A brief training session begins the week's work, stressing the nature of highly interactive learning and issues of group dynamics.
Discussions within the group produce better results than do single designers. A "script" records all design decisions. In our early work these scripts were paper documents, but now, thanks to the work of Bertrand Ibrahim at the University of Geneva, we have an online script editor, allowing the script to be entered in the computer and edited. In addition, materials are maintained in multiple languages. Static samples of such scripts are available at http://www.ics.uci.edu/~bork.
The third step is implementation, both for code and for media. From the script we move to a running program. Much of the code is written automatically from the computer-stored scripts. Media are developed by specialists in the area involved. The pieces are put together, and beta testing shows any problems, which we correct.
The last step in development is student evaluation, both formative and summative. This is an important activity for developing effective learning material. No matter how competent the designers are, the real test is how the learning units work with students. This is not the same as beta testing.
Several stages of formative evaluation and improvement should be conducted, with large numbers of students representative of the desired audience. Professional evaluators plan the details. The computer stores much of the information as the students progress through the material. We seek the parts of the program where learning is weak and improve these areas. Motivational factors are important in formative evaluation. Summative evaluation compares the new learning material with older approaches to learning, such as lecture-based methods. We are interested in major improvements in learning, not just small differences revealed by statistics. Papers describe the system in more detail.
ER: Is there a use for the Internet in these learning structures?
Bork: At present, I see little use for the Internet in the full delivery of learning. The current Internet, with overcrowding and limited bandwidth, does not encourage the delivery of large amounts of highly interactive materials. Newer broadband networks will probably be adequate. An examination of the "courses" offered on the Internet now, through web sites, shows mostly an information-delivery model of education. The overall quality is poor.
We can currently supply highly interactive programs by CD-ROM or DVD-ROM. In some cases, the Internet may be useful for providing supplementary web materials, much like a traditional library. Later, broadband networks may be used for delivery, but this will not be possible for a long time in poorer countries. Better organization and standards could increase the possibilities.
One immediate use of the Internet could be the storage and retrieval of the long-range records necessary for full interactivity.
ER: How will the use of technology-based learning models affect faculty?
Bork: Teaching faculty, in the sense that we know them today, may cease to exist, except for in smaller, advanced courses. But their skills and experiences will be important in the design of learning modules. This raises many issues about the structure of universities, such as the need to maintain research. If learning units are very profitable, they may support research!
ER: What is your view of the research conducted on the effectiveness of technology-based courses?
Bork: Research in learning is often poor. Since there are many variables affecting learning, most of them probably still unknown, we need to conduct such research with very large groups of students. Further, we need to give much more attention to practical research that improves learning.
ER: Do you recommend technology-based learning structures for use on campus or for use in distance learning?
Bork: At present, probably both. But soon the market will move toward distance learning. This assumes the existence of effective distance-learning courses. At the university, the place to begin with distance learning is in the large beginning courses, which account for about half of the university student load. Significant improvement in learning, at lower cost, is likely.
ER: Wouldn't the costs of developing highly interactive learning courses and delivering them to students be significant? What is your cost estimate per course?
Bork: It is a serious mistake to look at just the cost of development, as we should have learned from the United Kingdom Open University. The critical number is the cost per student for effective learning, including both development and delivery.
High costs of development can lead to low costs per student, if many students use the material. The development of a course at the Open University involves several million dollars. But its cost per student is less than half that of courses at conventional universities. See Sir John Daniel's recent book for more details about this.
We can do even better than this with highly interactive material. Our costs for development are about the same as the Open University costs. One way of viewing this is that it costs about $30,000 per student-hour of high-quality learning material. But we need to do the experiments to verify this. Experience with developing highly interactive software, and improvements in our software, may lead to lower costs for development. Our costs for delivery could be lower than those of the Open University. Our interactive materials might largely take over the role of the Open University's tutors, an expensive factor in its delivery system.
Costs per student are harder to estimate, since these costs depend on student numbers, difficult to determine without more experience. I do not know, for example, how many students study physics and calculus each year, but the number is large, because it includes all science and engineering students. So highly effective highly interactive distance-learning courses would have a large potential market, making them much cheaper per student than current courses, and if well developed, they will be much superior for almost all students. Bright students in high school could use these courses, perhaps instead of Advanced Placement physics and calculus, further reducing the cost per student. Versions of the courses in other languages could be marketed around the world and made available by organizations such as the World Bank or UNESCO in developing countries. Eventually, millions of students might use a highly interactive physics or calculus sequence. Similar considerations apply to other courses.
Another very important way to view this question is to consider what the costs will be to society worldwide if we do not develop better and more far-reaching learning. We cannot afford this!
ER: Are there ways to reduce the development costs for the college or university?
Bork: Development costs are lower for second-rate material. Universities are wasting, nationally and internationally, vast sums of money developing such poor material. The typical approach is to give some released time to faculty and to give limited support for programming and media production. It is unlikely, almost impossible, that good learning material will be developed this way.
Such development usually starts from a lecture-based course. University faculty often know no other way of conducting learning activities. Passive lecture material, or any other passive learning mode, is unlikely to lead to highly interactive individualized learning. These two types of learning represent incompatible models of how learning should take place.
ER: Do you see a role for government funding? commercial funding?
Bork: Presently neither is funding the development of major highly interactive computer-based learning units. Governments could fund such experimental efforts. Private foundations could also be important in these experiments, as could commercial funding.
Full development, after successful trials, would probably be done commercially, since there will be large profits to be made. New organizations, formed for the development of highly interactive learning modules, are likely to be the most successful because they do not start from earlier conceptions of what might happen.
Both public and private organizations already spend large amounts of money to improve learning, but not necessarily in the best directions.
ER: If private companies and the for-profit educational industry enter the market for interactive learning courses, how will colleges and universities compete?
Bork: Peter Drucker predicted that universities will die in thirty years because universities will not compete successfully with companies. A few may, but not with current strategies. Small wealthy colleges will probably survive.
Universities are too stuck in their current ways of doing things to be able to compete with well-developed material from "outside." Most university faculty and administrators do not appreciate the current problems of learning and so are not prepared for these future directions.
This leaves many open issues, beyond this discussion. Universities serve many functions other than learning. For example, as I mentioned, there is the question of the future home of research.
ER: How do we begin?
Bork: DO THE EXPERIMENT! Develop a group of highly interactive learning sequences, and evaluate them carefully. Then we can move forward from sound evidence.