INTRODUCTION
GENERATIVE CAI
In the 1960's, researchers created a number of Computer Assisted
Instructional (CAI) systems that were generative
(Uhr, 1969). These programs generated sets of problems designed to enhance
student performance in skill-based domains, primarily arithmetic and vocabulary
recall. Essentially, these were automated flash card systems, designed to
present the student with a problem, receive and record the student's response,
and tabulate the student's overall performance on the task
THE STRUCTURE OF AN ITS SYSTEM
Intelligent tutoring systems consist of four
different subsystems or modules: the interface module, the expert module, the
student module, and the tutor module. The interface module provides the means for the student to interact
with the ITS, usually through a graphical user interface and sometimes
through a rich simulation of the task domain the student is learning
(e.g., controlling a power plant or performing a medical operation). The expert module references an expert or
domain model containing a description of the knowledge or behaviors that
represent expertise in the subject-matter domain the ITS is teaching, often an expert
system or cognitive model. An example would be the kind of
diagnostic and subsequent corrective actions an expert technician takes when
confronted with a malfunctioning thermostat. The student module uses a student model containing descriptions of
student knowledge or behaviors, including his misconceptions and knowledge
gaps. An apprentice technician might, for instance, believe a thermostat also
signals too high temperatures to a furnace (misconception) or might not know
about thermostats that also gauge the outdoor temperature (knowledge gap). A
mismatch between a student's behavior or knowledge and the expert's presumed
behavior or knowledge is signaled to the tutor
module, which subsequently takes corrective action, such as providing
feedback or remedial instruction. To be able to do this, it needs information
about what a human tutor in such situations would do: the tutor model.
An intelligent tutoring system is only as
effective as the various models it relies on to adequately model expert,
student and tutor knowledge and behavior. Thus, building an ITS needs careful
preparation in terms of describing the knowledge and possible behaviors of
experts, students and tutors. This description needs to be done in a formal
language in order that the ITS may process the information and draw
inferences in order to generate feedback or instruction. Therefore a mere
description is not enough; the knowledge contained in the models should be
organized and linked to an inference
engine. It is through the latter's interaction with the descriptive data
that tutorial feedback is generated.
USE IN PRACTICE
All this is a substantial amount of work, even if authoring
tools have become available to ease the task. This means that building an
ITS is an option only in situations in which they, in spite of their relatively
high development costs, still reduce the overall costs through reducing the
need for human instructors or sufficiently boosting overall productivity. Such
situations occur when large groups need to be tutored simultaneously or many
replicated tutoring efforts are needed. Cases in point are technical training
situations such as training of military recruits and high school mathematics.
One specific type of intelligent tutoring system, cognitive
tutors, has been incorporated into mathematics curricula in a substantial
number of United States
high schools, producing improved student learning outcomes on final exams and
standardized tests. Intelligent tutoring systems have been constructed to help
students learn geography, circuits, medical diagnosis, computer programming,
mathematics, physics, genetics, chemistry, etc.
CHALLENGES TO ITS
By the
mid-1980's, much of enthusiasm in AI for creating "thinking"
computers had waned as the field began to mature. Researchers turned to the
more prosaic tasks of building expert systems that could function well in
constrained domains, such as troubleshooting and diagnostic systems. At the
same time, as ITS began to move out of the AI laboratories into classrooms and
other instructional settings, they began to attract critical reactions. Some
shortcomings of ITS became apparent as researchers realized that the problems
associated with creating ITS were more intractable than they had originally
anticipated. Rosenberg
notes that most papers about ITS make few references to the education
literature; the majority are grounded in the computing literature. He asserts
that much ITS work suffers from two major flaws:
The systems are
not grounded in a substantiated model of learning. Model formulation should be
preceded by protocol analysis, but very little analysis is done, almost none of
it qualitative. ITS models should be validated by the teachers and students who
will use the systems, but ITS researchers do not appear to consult these
experts.
Testing is
incomplete, inconclusive, or in some cases totally lacking. Data on
computerized tutorials are, at best, mixed. The almost universally positive
claims for ITS and other computerized instructional systems, most notable in
the education literature, are based on results from severely flawed tests.
It was obvious
that the basic premises of ITS research needed revision.
Intelligent tutoring systems, four different subsystems
or modules:
1. The interface module: It provides the means for the student to
interact with the ITS, usually through a graphical user
interface and sometimes through a rich simulation of the task domain the student is
learning (e.g., controlling a power plant or performing a medical operation).
2. The expert module: It references an expert or domain model
containing a description of the knowledge or behaviors that represent expertise
in the subject-matter domain the ITS is teaching, often an expert system or cognitive model. An example would be the kind of
diagnostic and subsequent corrective actions an expert technician takes when
confronted with a malfunctioning thermostat.
3.The student module: It uses a student model containing descriptions
of student knowledge or behaviors, including his misconceptions and knowledge
gaps. An apprentice technician might, for instance, believe a thermostat also
signals too high temperatures to a furnace (misconception) or might not know
about thermostats that also gauge the outdoor temperature (knowledge gap)
4. The tutor module: A mismatch between a student's behavior or
knowledge and the expert's presumed behavior or knowledge is signaled to the tutor module, which subsequently
takes corrective action, such as providing feedback or remedial instruction. To
be able to do this, it needs information about what a human tutor in such
situations would do: the tutor model.
THE INTELLIGENT TUTORING SYSTEMS CONFERENCE
Ø in Montréal (Canada) by
Claude Frasson and Gilles Gauthier in 1988, 1992, 1996 and 2000;
Ø in San Antonio (US) by Carol Redfield and
Valerie Shute in 1998;
Ø in Biarritz
(France) and San Sebastian (Spain) by Guy Gouardères and Stefano
Cerri in 2002;
Ø in Maceio (Brazil) by Rosa
Maria Vicari and Fábio Paraguaçu in 2004;
Ø in Jhongli (Taiwan) by Tak-Wai Chan in 2006.
Ø The conference was recently
back in Montreal
in 2008 (for its 20th anniversary) by Roger Nkambou and Susanne Lajoie.
Ø ITS'2010 will be held in
Pittsburgh (US) by Jack Mostow, Judy Kay, and Vincent Aleven.
SOME INTELLIGENT TUTORING SYSTEMS ARE
Ø
The Love Machine
Ø
Toward Empathetic Agents in Tutoring Systems
Ø
Intelligent Tutoring Systems with Conversational Dialogue.
Ø
Using technology for learning &
teaching science. (as astronomy,
space research, physics, mathematics and the earth sciences. )
Ø
Pitch-perfect PC
Ø
The F-16 Maintenance Skills Tutor
Ø
City pushes computer tutor for
struggling algebra students.
PROSPECTS FOR CREATING AN INTELLIGENT TUTORING SYSTEM
Ø
Modularize the
curriculum.
Ø
Customize it for
different student populations.
Ø
Individualize the
presentation and assessment of the content.
Ø
Collect data which
instructors could use to tutor and remediate students.
The
first step in this process is to understand what others had done before us and
the implications for future developments.
ARTIFICIAL INTELLIGENCE AS TUTOR
In 1998, chemist Benny Johnson founded Quantum Simulations,
Inc. with high school mentor Dale Holder and colleague Rebecca Renshaw to
create highly interactive tutoring software for the sciences.
Student modeling remains at the core of ITS research (Holt,
Dubs, Jones & Greer, 1994). What distinguishes ITS from CAI is the goal of
being able to respond to the individual student's learning style to deliver
customized instruction.
DIAGNOSIS
Diagnosis means
that an ITS infers information about the learner's state on three levels.
At the
behavioral level, ignoring the
learner's knowledge and focusing only on the observable behavior.
At the
epistemic level, dealing with
the learner's knowledge state and attempting to infer that state based on
observed behavior.
At the
individual level, covering such
areas as the learner's personality, motivational style, self-concept in
relation to the domain in question, and conceptions the learner has of the ITS.
Wenger notes that, up to now, ITS have not been concerned with the individual
level. However, he advocates further research in this area as a prerequisite
for viewing the student as an active learner, rather than as a passive
recipient of knowledge
No comments:
Post a Comment