Smithtown: An Intelligent Tutoring System
This article describes Smithtown, one of a family of new instructional aids known as intelligent tutoring systems (ITSs). It employs artificial-intelligence methods to assist students in beginning-economics courses to improve their problem-solving skills.
Ther is no need to dwell on the importance of a sound understanding of economics for intelligent dealings in a complex modern society. Extremely heavy college enrollments in introductory economics courses, year-in and year-out, tend to belie the difficulties of most students in understanding this subject. Smithtown is an interesting development because it gives instructors new insights into student learning processes while at the same time helping students to more enjoyably master the principles of economic reasoning.
Intelligent tutoring systems like Smithtown increase students' involvement in the learning process of "interrogation and confrontation."
Up to now, most of the work with ITSs has been done in the controlled environments of learning laboratories with elaborate workstations too expensive to install on a large scale. The PC version that we outline here evolves from this research but has been specifically adapted to more affordable micro-computers for classroom use. The first section of the article discusses some of the underlying rationale of the ITS approach. Succeeding sections describe the structure of the PC version (written in object-oriented Smalltalk/V) and the future directions prompted by its promising results.
Smithtown aims to help students master the basic principles of microeconomics and understand how scientific reasonng can help them to become better observers and interpreters of economic behavior.
The program was initially developed at the University of Pittsburgh Learning Research & Development Center (LRDC) by Valerie Shute, Robert Glaser and Kalyani Raghavan, and coded in object-oriented LOOPS as a research tool for studying induction and the discovery behaviors of students exploring a new domain. The development had as a main objective the gaining of new insights into the acquisition of skills such as: being aware of the implications of varying one variable while keeping others constant; collecting baseline data to observe change; and collecting sufficient information to formulate and test hypotheses.
It was recognized at the outset, however, that the project had significant potential for expanding subject-matter learning as well as improving discovery skills through the practice of problem-solving and computer-implemented diagnoses of learning problems. The program, as it has evolved, continues to perform this dual function--as an important tool for studying learning processes and as a help to students in overcoming learning difficulties.
In addition to being more research oriented, an ITS differs from traditional computer-assisted instruction (CAI) programs in not being as highly structured. Whereas CAI programs tend to be organized around decision trees, an ITS allows for more open-ended programs.
The Smithtown ITS, for example, provides the student with a set of discovery tools and invites the performance and analysis of self-directed experiments. These can be interrupted at any point to test hypotheses that evaluate understanding. The program uses artificial-intelligence techniques to monitor students' reasoning, to track errors to their source and, based on the diagnoses, to offer advice on strengthening problem-solving skills. The learning of the subject matter happens incidentally as a byproduct of the more effective use of discovery tools.
The Program Structure
The PC adaptation to be described is less elaborate than the LRDC version of SMithtown. However, it contains most of the key elements. The program begins by situating students in the imaginary community of Smithtown, where they may manipulate outcomes in a series of consumer markets by changing the prices of commodities or other environmental factors such as population, per capita income or number of suppliers. …