Expert Systems in Education and Training: Automated Job Aids or Sophisticated

Instructional Media?
Author(s): Alexander J. Romiszowski
Source: Educational Technology , October 1987, Vol. 27, No. 10, Critical Issues in
Educational Technology (October 1987), pp. 22-30
Published by: Educational Technology Publications, Inc.

Stable URL: https://www.jstor.org/stable/44427248

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

is collaborating with JSTOR to digitize, preserve and extend access to Educational Technology

This content downloaded from

154.120.242.82 on Wed, 08 Jun 2022 08:29:51 UTC
All use subject to https://about.jstor.org/terms
 Of the three main areas of Artificial Intelli-
Expert Systems in gence (Al) research, referred to above, namely
Education and Training: Natural Language Processing, Robotics, and
Expert Systems, the third is the area in which,
Automated Job Aids or suddenly, there seems to be real promise of com-
mercially viable large-scale application. Perhaps
Sophisticated for this reason, more than any other, Expert
Instructional Media? Systems have caught the attention of educators,
especially the "high-tech" fringe. Perhaps for
this reason, aļso, reactionary voices have been
raised, questioning the educational utility, eco-
nomic viability, or indeed relevance to educa-
tion, of Expert Systems.
Alexander J. Romiszowski
On the one hand, Al enthusiasts point to the
very rapid growth of interest and development
Where Are We? Fact and Fiction effort in the field among business. Newquist
of Expert Systems (1986) claims that about half the Fortune 500
companies are already investing in the develop-
Every month, the educational and training journals
ment of expert systems. Their use in education
publish an ever growing stream of articles on the
is but a question of time. Wilson and Welsh (1986)
topic of Artificial Intelligence and particularly Ex-
refer to knowledge-based expert systems as "some-
pert Systems. Why the sudden interest in a relative-
ly slowly developing field of research, which has
thing new under the (education and training)
sun," emphasizing the very rapid growth in the
"been around" so to speak for some thirty years
and has been more noted for its failure to deliver methodology for developing such systems and the
proliferation of software already available at
promised results on time than for real progress in
economic prices for use on personal computers,
terms of successful practical applications? Some
that may enable any educatorio become involved.
twenty years ago, we were "on the verge" of
But they do not really clarify WHY an educator
automatic language translation by computer.
should wish to get involved.
Today, despite significant progress in natural On the other hand, performance technologists
language interpretation and programming, the goalsuch as Harmon (1986) and Kearsley (1986a),
of a practical automated interpreter remains out see expert systems, especially the small and
of reach. Despite much progress in industrial
relatively simple knowledge-based "advisors," as
robotics (little more than rather sophisticated
just "a new medium for developing complex job-
numerically programmed machinęry) we are a
aids." This would seem to limit their application
long way from constructing the type of general
to the non-training aspects of real-life performance
purpose robot exemplified by Hal in "2001 : A
problem-solving in industry and, in education, to
Space Odyssey," or indeed any of the "intelli-
the role of labor-saving devices, like slide-rules,
gent" automated butlers that we see so often
pocket calculators, word processors, and spread-
in television programs set "just a little into the
sheets. But, as expert systems are job aids to as-
future." Finally, we aré yet a long way indeed
sist with the solution of "complex" problems,
from constructing the all-purpose reasoning ma-
which require the use of a modicum of "intelli-
chine that could pass the Turing test of intelli-
gence," is it in fact a "good thing" to supply
gence (machine and human decision-making be-
students with a job aid? Will it stop students
ing indistinguishable).
from thinking for themselves? Should expert
It is in this latter field, however, that somesystems be banned from education?
limited progress of a significant and practical Yet other authors point out that at this time, the
nature has recently been achieved: in the de- technology is only capable of dealing with very
velopment and economic application of Expert tightly defined problem-solving tasks and, although
Systems that emulate human expertise in small, some of the very big (and very expensive) expert
tightly defined and highly structured domainssystems (such as MYCIN) do emulate world-class
of complex problem-solving. expertise in a difficult problem area (Harmon and
King, 1985), the smaller systems, incorporating
knowledge bases of some 20, 50, or 100 rules, are
Alexander J. Romiszowski is Professor, Instructional De-no more than glorified algorithms that can be (and
sign, Development, and Evaluation, Syracuse University,indeed were) developed by non-computer tech-
Syracuse, New York. niques, such as decision-tables (Pipe, 1986).

22 EDUCATIONAL TECHNOLOGY/October 1987

This content downloaded from

154.120.242.82 on Wed, 08 Jun 2022 08:29:51 UTC
All use subject to https://about.jstor.org/terms
 All these viewpoints represent rather extreme design, development, and use of algorithms, logical
positions, all contain a modicum of truth, but trees, and decision tables as job aids that reduced
none really pinpoint the true potential, or indeed the training requirement (Lewis et al., 1967:
the true limitations, of knowledge-based expert Lewis and Wolfenden 1969; Lewis 1970; Wheatley
systems for education. Knowledge-based expert and Unwin 1972). It was only natural that these
systems, even those developed on relatively cheap two developments should be fused and popular-
and low-power software (for example, Texas ized. Thus, in 1970, the book Analyzing Perfor-
Instruments' PERSONAL CONSULTANT, 1985) mance Problems (Mager and Pipe, 1970) was pub-
can handle 200 or more interdependent rules, even lished. This explained, with typical Mager-style
"fuzzy" probabilistic relations, with differing clarity and with bags of practical examples, the
levels of confidence and can supply alternative principles of analysis and solution of job-perfor-
solutions with different resultant levels of proba- mance problems (including the use of algorithms
bility. Even simple algorithmic job-aids can lead to as job aids) and provided the reader with a job aid
effective learning, and expert systems offer much in the form of an algorithm (see Figure 1 ).
more potential for productive learning. On the Now, unlike most of the applications of this
other hand, they are not a panacea forali learning
technique developed by Lewis and his collabor-
problems- what is? ators, which were "true" algorithms, the flow-
chart in Figure 1 is a "quasi-algorithm." A true
Where Have We Come From? algorithm is "a procedure that, if followed cor-
The Evolution of the Job Aid rectly, all users solve all the problems in a domain
Expert systems, in their intended practical ap-
correctly and in the same way" (Landa, 1976).
plications, are indeed, job aids (or to use the moreExamples of true algorithms are procedures for
correct term, job PERFORMANCE aids). They the division of two numbers, for the calculation of
are ". . . . software programs that HELP YOUan annual income tax return (given a set of earn-
SOLVE complex reasoning tasks that normally
ings and claims data), and so on. On the other
hand, the procedures for analyzing a complex
require an expert . . ." (Texas Instruments, 1985).
The expert system will help a novice or relatively
problem in order to decide how to solve it (what
inexperienced problem solver to match the skilltoofdivide by what), or to decide how to enter
acknowledged experts in the domain in question. and represent a given taxpayer's earnings and
The very large expert systems that emulate world-claims, are seldom truly algorithmic. There is
class expertise in a domain that is so complex that
usually some degree of value judgment, weighing
few humans ever qualify as "experts" may con- of viable alternatives, or decision-making on the
tain many hundreds, even thousands of separate, basis of incomplete data involved. For example,
but interrelated rules (and therefore take thou- in Figure 1, the second question lozenge- "skill
sands of manweeks and millions of dollars to de- deficiency?"- demands a complex analytical proc-
velop-see Wilson and Welsh, 1986). The more ess of observing performers and perhaps inter-
modest systems, however, may emulate only rea- viewing them, as well as an appropriately formed
sonable levels of expertise in a domain where concept of "skill" on the part of the observer/
many humans actually reach proficiency, but interviewer. No wonder that in the same case
where many more need to take decisions and may study simulation, trainee performance technolo-
need advice. The first category may replace a verygists often disagree as to whether the correct
rare and expensive human expert. The second answer is YES or NO. Even experienced perfor-
category rarely replace, but rather ASSIST the mance technologists disagree sometimes. The
user in the search for an optimum solution to a "skill deficiency" decision point, as indeed most
complex problem. of the others in Figure 1, involves a complex
Let us look at the history of job-performance of subordinate "if-then" decisions that the "ex-
aids for decision-making tasks, to see where these
pert" performance analyst learns to apply through
"expert advisors" fit in. The impetus to the useexperience. The job aid in Figure 1 is useful, but
of decision-making job-performance-aids came in only insofar as it reminds the analyst to ASK all
the mid-1960s, from the work of early perfor- the relevant questions in an efficient SEQUENCE.
mance technologists, such as Gilbert (1967), It does little to help him or her ANSWER the
Harless (1968), and the work of US-based con- questions- for this it was necessary to write the
sultancy organizations such as Geary Rummler's whole book of rules and illustrate with adequate
Praxis Corporation. On the other side of the examples and, in practice, then follow up with
Atlantic, in England, the Cambridge Consultants more practical examples in simulated case study
group was developing the "art" of flowcharting situations.
into a powerful technology for the systematic A further example of a "quasi-algorithm" is

EDUCATIONAL TECHNOLOGY/October 1987 23

This content downloaded from

154.120.242.82 on Wed, 08 Jun 2022 08:29:51 UTC
All use subject to https://about.jstor.org/terms
 Are
I Describe ' Does„ , I» object important 15
I performance J topic deal , yes too large no aspects of no the
' deficiency J with concrete - +- or too small to - •- the object object
observable be easily hidden or available?
objects? observed? obscured?

no 'yes / yes no yes

' / y' Why not

u real
' /yS use the
object?

O.^ Yesy
Does
• j i Does
topic • deal j i c Consider
with concepts yes require a Ves using a
not themselves

observable but , mnck-un

... • . presentation? , mnck-unmocK
mocK UP
UP
S Knowledge N. with ... observable .
Yes S or ' effects?
I >sSkill deficiency? yS ./V
no ļ no
X. No| w / Performance n. Yes Does A motion
N. N. punishing? S I
the topic film or
ļ
require that yes working
students learn

y / Remove to recognise or live

punishment
or copy demonstra-
Arrange XYes motions? tion
formal ¥ jS N.
training no

N. ^ / Non-performance yes
S' >v '. rewarding? yS ļ Visual media Still visual
No / Performs >v ^
' often? / /
probably not media are
necessary good enough
^ ^ '. / ^ Arrange
/v consequence
Arrange Yes

practice

w / Perfo
>v matters? yS ļ
Arrange
feedback
_ >v yS Figure 2. Example of part of a "Quas
assist the task of media selection (Rom
consequence
1974).
w / Obstacles to yes
w * ' good performance / |

Remove need to follow several paths at once (Romiszowski,

1970). In Figure 2, for example, a topic may deal
obstacles

with BOTH concrete observable objects AND

abstract concepts (an airplane's wing); and in
Figure 1 , an analyst following the flowchart may
correctly diagnose a deficiency in the skills of a
given group of performers and may set about the
Figure 1. Example of a "Quasi-Algorithm" to assist the
development and implementation of a costly train-
task of performance problem analysis and solution.
ing program, only to find later on that there are a
series of factors in the job environment- inbuilt
punishments and obstacles- that are beyond his
presented in Figure 2. This is part of a setpower of to influence and that totally undermine
flowcharts developed by the author (Romiszow- even the most efficient training effort (e.g., the
ski, 1968, 1974) to act as instructors' job aids civil service career structure and its negative in-
for the selection of instructional media. Experi- fluence on job motivation).
ence in using these with many groups of instruc- During the 1970s, therefore, there was a trend
tors led to the observation that although theaway use from the use of algorithmic job aids, except
of job-aid flowcharts significantly improved in the case of true algorithmic procedures. Job
lesson planning decisions, SOME instructors aids werefor quasi-algorithmic, part-heuristic decision-
SOMETIMES misled by the very form of the flow- making procedures tended towards the use of IF-
charts in that the YES/NO decision format was THEN tables and WHIF (WHen and/or IF) charts
taken to always imply a categorical one-way route (Horabin, 1971; Horn, 1974) and Cognitive Sche-
through the decision process. But due to the mata (Romiszowski, 1980, 1981). IF-THEN and
quasi-algorithmic nature of the task, one may WHIF charts are a step towards the listing of the

24 EDUCATIONAL TECHNOLOGY/October 1987

This content downloaded from

154.120.242.82 on Wed, 08 Jun 2022 08:29:51 UTC
All use subject to https://about.jstor.org/terms
 ' ofcJĚCTives / X.

/'Zrrfìì*/''"0™ NeV" 'V^^JfE>'

fS_///¿V '%%sV^5i' / 7Y/'h '' '
I II - V-4
fil*N^1 £ ¿Í [mummy 'pe*fo*n' W fí¿r '
b^1iií/jk/°^¿r fil*N^1 £ ¿Í [mummy m 'pe*fo*n' wfŁŁ W i- fí¿r va
I Oř I 1 1 S ' WHAT V PCKFOĄMAHCI J ^ Ig |Ia£-o,J
W#n'il ' APf rue '. y nm ļ §Ę§JrHe / I ' V SEL£Cn0N J I ^
Wfa}j^ò?MIT^T-'^ra" /fifJSSr ' /
ww^' wMNiy^Ą^^i
Y/$xs£ /Ž? / ' ' W
X
CP^ / /

xVvv/fV^L -
,w /^fT~ "Pl^v řAClLmti
^gSfsS 1
Figure 3. An Analysis schema for performance problems:
Figure 4. Factors affecting t
Questions to ask, answers to compare and solution- struction: a conceptual sche
components to select (Romiszowski, 1980, 1981). their Interrelation (Romiszow

separate but often interrelated rules that underly the need to build up exp
complex decisions- a step towards the develop- experience and training.
ment of knowledge bases for expert systems.
Cognitive schemata were an attempt to repre- It is in this area of quas
heuristic decision-making that expert systems
sent the structure and interrelatedness of a set
of factors that should, together, be taken intoshould have ONE of their principal impacts on edu-
cation and training. The type of decision-making
consideration in complex decision-making. Figures
tasks described in the two examples above, al-
3 and 4 are two such cognitive schemata developed
though not as precisely defined as some, are
by the author for the same areas of decision-mak-
amenable to the development of REASONABLY
ing represented by the flowcharts in Figures 1 and
expert systems, capable of diagnosis and decisions
2. The circular nature was intended to emphasize
the integrated nature of the decision-makingequivalent to most successful instructional de-
signers, or performance technologists. There are
process. As job aids, they succeeded in overcom-
ing the "blinkered" one-route decision-making precedents for this belief. Kearsley (1986b) has
which algorithmic flowcharts tend to promote, developed two working knowledge-based systems,
but apart from that they are no real advance- which assist the trainer to take cost-benefit analysis
decisions in relation to training programs and
they also only act as memory aids to the factors
that should be considered, but do not aid much assess the suitability of a given course or topic for
with HOW to decide. computer-based instruction. Under military con-
An exhaustive set of IF-THEN rules set out in a tract, Singer and Perez (1986) have successfully
chart are more helpful in this respect. But, again,
developed a first version of an expert system for
if the number of interrelated rules is large, and
the design of training devices for military train-
the decision-maker is not experienced, there is a applications, an area of application which in-
ing
limit to the number of separate rules that a human
volves an annual budget of over seven billion dol-
being can keep in short-term memory and con- lars a year. The present author is currently work-
ing
sider in relation to each other- probably related to to develop experimental systems in the two
skill areas exemplified above- performance prob-
Miller's (1967) "magic number 7 . . . plus or minus
2." Once more, we cannot seem to get away from lem solution and media selection.

EDUCATIONAL TECHNOLOGY/October 1987 25

This content downloaded from

154.120.242.82 on Wed, 08 Jun 2022 08:29:51 UTC
All use subject to https://about.jstor.org/terms
 »»^COMPUTER J

TUTOR TOOL TUTEE

USER:

Intelligent Commercially Use of LISP or

computer-assisted produced expert PROLOG in schools.
instruction (ICAI) systems used to

Intelligent data- (JOB AID) struct expert

base search systems for the
STUDENT assistant. Student learns by domain under
(EMBEDDED TRAINING) repeated use. study.

Intelligent Stand-alone expert Teacher/SME

EMBEDDED training systems on I.D. constructs expert
TEACHER/IN- 'n ^00' aPP''" decision making, systems on
STRUCTION cations software. planning, con- specific domains
DESIGNER/
AUTHOR Intelligent - 1 and 2.
teacher-training Intelligent
systems. authoring systems
for CAI.

Figure 5. The "Field" of A.I. applications in eduation and training (Romiszowski

Where Are We Going? we be teachers, instructors, instructional design-

The Educational Benefits of Experters or authors/developers of instructional ma-
Systems
Readers with an interest in formal education,
terials).
rather than training, may be excused for wonder-
ing if this article was written for them. Expert Systems/AI as "TUTORS"
Hopefully,
In
in this section, they will see that it is of relevance the left-most column of our schema, we
to the loftier aims of a general education- and have "Intelligent Tutoring Systems" or "Intel-
how! ligent Computer-Assisted Instruction" (ICAI).
Elsewhere (Romiszowski, 1986; 1987, in press) These are custom-designed, special-purpose, tu-
I have used Taylor's (1980) schema of computers torial packages such as those described by Slee-
in education, "Tutor, Tool, Tutee," as an organiz- man and Brown (1982), Carbonell (1970), Pask
ing framework for the analysis of potential Al (1984), and Park and Seidel (1986), the last
applications in education. The overall schema is authors making an interesting comparison be-
reproduced in Figure 5. We shall not examine all tween the roots, the methodologies, and the
the entries in all the cells in full detail, as this results of "conventional" CAI and the new, "in-
would be repeating earlier works. Instead, we telligent" ICAI.
shall single out the applications of expert systems Intelligent tutoring systems are more complex
that seem, in this author's opinion, to offer the in basic structure than expert systems, being in a
greatest potential for educational benefits at not sense a combination of subject matter expert,
too excessive cost. student-results-analysis-and-interpretation expert,
The reader will note that Taylor's three cate- as well as expert tutor. Being extra complex and
gories of "role" that the computer may adopt, extra specialized, they are extra costly to develop.
have in my schema been crossed with two cate- This alone might rule them out as practical solu-
gories of user- our students (whether theytions be to formal educational problems. Coupled to
young children in school, university students, or however, is the almost total absence in the
this,
trainees in business) and "ourselves" (whether literature of any evidence that this high cost re-

26 EDUCATIONAL TECHNOLOGY/October 1 987

This content downloaded from

154.120.242.82 on Wed, 08 Jun 2022 08:29:51 UTC
All use subject to https://about.jstor.org/terms
suits in any benefit. Of the 15 or so intelligent rules and a sequence and applies them to exam-
tutorial systems which according to the litera- ples- RUL-EG), using an expert system as a job
ture seem to have reached the stage of full develop- aid involves working on a series of specific exam-
ment, only about five have been used regularly ples of problem solving and, perhaps discovering
on anything like a large scale and of these, not in time, not only the individual rules that the
one seems to have been systematically evaluated. system uses to solve the problems, but the inter-
Let us leave the TUTOR mode, therefore, and relationships that exist between them.
concentrate our analysis on the TOOL and TUTEE In the hands of a skillful teacher, an appropriate
columns of the schema. Here we seem to get a very expert system, used initially as a problem-solving
different picture of educational promise and cost- job aid, may become a medium for the discovery
benefit ratio.
learning by the student of the original experts'
cognitive structures that are built into the system's
Expert Systems as "TOOLS" knowledge base. Far from being the threat to the
In the TOOL category, we have, at the top, use student's independent thinking processes, hinted
by students of existing knowledge bases and expert at in our opening paragraphs, expert systems in
systems as, firstly, job aids to assist in the solution the classroom may be used to promote deep
of relevant categories of problems. For example, analytical and productive thinking, through a
an economics course might incorporate an expert process of guided discovery.
"stockbroker" system, an engineering course will
use an existing "bridge designer" system, a high In the lower part of the TOOL category, we
school class may tune into an existing "weather have the use of expert systems as, principally,
forecasting" system as part of a geography unit and instructional design and development tools, by
an elementary group may use an expert advisor teachers and designers. We have already analyzed
on "productive writing." All these, except perhaps some existing and proposed stand-alone expert
the last, may be commercial expert systems, de- systems for individual tasks like cost-benefit
veloped for real-time use: their application in edu- analysis, media selection, or simulator design.
cation is a spin-off benefit, at reasonable cost once The armed forces are, as a whole, convinced that
such tools are essential to maintain both the
the systems are in widespread use.
The main spin-off benefit, however, is from the quantity and the quality of instructional design
learning processes that may be associated with the decisions in military training. Any progress regis-
use of these job aids. When using a job aid fre- tered here will, no doubt, seep through to educa-
quently enough, the user inevitably learns much, tional decision-making in time. The other signifi-
if not all of the procedure involved. In the case of cant movement in this field is the development of
algorithmic procedures, this learning is mere embedded expert system tools within CAI author-
memorization. In the case of general-purpose ing systems (Merrill, 1 985 ; Merrill and Wood, 1984;
algorithms, which have some grounding in theory McAleese, 1 986). This development is seen by edu-
(arithmetical procedures, for example, as opposed cators as essential to ensure quality in computer-
to one-purpose algorithms like the procedure to based courseware design. It may be that intel-
operate a particular photocopier) it is desirable to ligent tutoring systems are a thing of the dis-
get away from mere memorization by getting the tant future, but intelligent authoring systems
learners to derive their own algorithms from "first may indeed be instrumental in saving our "con-
principles" (Landa, 1976). But in the case of learn- ventional" CAI systems from extinction due to
ing from expert systems, the situation is much self-destructive mediocrity (Romiszowski, 1986).
more interesting from a learning theory viewpoint.
In the first place, most respectable expert system Expert Systems as "TUTEES"
software shells provide the means to include ex- Turning now to the TUTEE category, we
planatory messages. The user may, whenever encounter another potential development and one
asked to furnish data, inquire WHY that particular about which I am personally very excited. This is
information is required for the problem. And when the use of simple expert system "shells" to enable
the expert system presents a decision or solution, students in school to develop their own expert
the user may ask HOW that decision was reached, systems on topics that they are studying. This
where upon the system retraces the logic of its combines a whole series of pedagogically attractive
decision-making process. Thus, the user not only aspects in a practical and motivational new type of
gets a solution, but a full explanation. And all exercise:
this is restricted to a series of SPECIFIC EXAM- • Learning by doing and in the process
PLES of problems. Unlike the case of using an creating new knowledge, or at least new
algorithm as a job aid (the user receives a set of structure for the knowledge found in books

EDUCATIONAL TECHNOLOGY/October 1987 27

This content downloaded from

154.120.242.82 on Wed, 08 Jun 2022 08:29:51 UTC
All use subject to https://about.jstor.org/terms
 and other sources, as well as working job- technical reports of validation studies. THAT
aids that may be useful later. STRUCTURING HAS SHOWN US WHERE
ANSWERS ARE NEEDED AND WHAT
• Learning by discovery, in the domain of
VARIABLES ARE INVOLVED IN OBTAIN-
higher-order problem-solving and the crea-
ING THE ANSWERS.
tion of cognitive strategies and theories.
• Applying a logic of the structure of knowl-
Will We Get There? When?
edge to segments of one's own store of
To take the last question first, many of the po-
knowledge in order to create a computer-
tential benefits of using expert systems in educa-
ized knowledge-base and thus learning to
tion and training are attainable now. Commercial
reflect on and recognize the rest of one's
knowledge bases and expert systems are beginning
personal knowledge-base.
to appear and will do so at an ever-increasing rate.
An interesting application of this approach is
Initially, they will be hard to come by, due to the
reported by Lippert and Trollip (1986), using a
cost of their development and the commercial
specially constructed, simplified expert system
value of the embedded expertise to the developer
shell with 12- to 15-year-old students on a variety
and client. But some public-domain and subscrip-
of domains of problem-solving.
tion service systems are already in development,
Other researchers have been working along
and these will serve to start the ball rolling in the
similar lines with older age-groups. Starf ield
TOOLS department. Tools for teachers and in-
et al. (1983) have been using the approach for
structional designers are also in the pipeline, but
some time with engineering students.
their ready availability may be somewhat delayed
The use of expert system software in this way
by the normal budget restrictions that exist in
is, in its intent, very similar to the use of LOGO education.
in schools (Papert, 1980). Apart from the direct The TUTEE applications depend on the avail-
learning that takes place, associated to the CON-
ability of simple, easy-to-use knowledge-base
TENT which is being processed (geometry in
building software. Some of the commercially
turtle LOGO and almost any well structured set
available packages may be suitable for use by
of principles in expert system development), the
university students, but even in these cases the
students learn "powerful ideas" of general applic-
learning time involved in mastering the software
ability; in the latter case, associated with general
may be excessive. Research projects like the ones
approaches to problem formulation and problem described earlier have in the main found it nec-
solving.
essary to develop special, much simplified soft-
Finally, there may be the one other important
ware, especially when young school children are
development, analogous to students constructing
to be the budding "knowledge engineers." These
expert systems to structure their knowledge-base
simplified shells will no doubt soon become
on a specific subject domain. This is the possible
generally available, and should open the way for
use of ex pert system software, in the TUTEE mode,
many other teachers to follow. Given a suitable,
by teachers, instructional designers and, especially,
the authors of books and other educational materi-
easy-to-learn shell, the rest depends only on the
als.
imagination of the teacher to choose suitable
knowledge-base building projects and on his or her
To illustrate this point and to conclude this
instructional skills to ensure that the students get
section, I quote from the final paragraphs of the
the most out of the experience.
paper of Singer and Perez (1986), when describ-
Returning now to the first question- will edu-
ing their project to design an expert system for the
cators reap the potential benefits we- have out-
design of training devices:
lined?- I would hesitate to make a guess. Apart
With the use of a knowledge engineering from the flavor of futurology in this article, and
tool such as KES, we have SUCCESSFULLY the plethora of perhaps shaky hypotheses derived
structured what is known about the character- from a few as yet unfinished studies, the other
istics and instructional features of training great unknown is the general posture that the
teaching profession will adopt to Al's inroads into
devices. As researchers in training and train-
ing devices, we have ACQUIRED SOME the traditional preserves of the teacher. At the
KNOWLEDGE about the design of training 1986 ADCIS International Conference, Peter Dean
devices. We believe that this process CAN BE of IBM addressed a number of the underlying, but
APPLIED TO OTHER FUZZILY BOUNDED in his opinion perhaps mistaken, assumptions made
AND NON WELL-DEFINED KNOWLEDGE
by the users and the practitioners of computer-
DOMAINS as well. The process has FORCED based education. Among these were:
US TO STRUCTURE THE KNOWLEDGE
AVAILABLE from field experiments and • Teachers will never be replaced by com-

28 EDUCATIONAL TECHNOLOGY/October 1 987

This content downloaded from

154.120.242.82 on Wed, 08 Jun 2022 08:29:51 UTC
All use subject to https://about.jstor.org/terms
 puters (in his view, they will be partly Landa, L.N. Instructional Regulation and Control: Cyber-
replaced and, thus, their role will change netics, Algorithmization, and Heuristics in Educationě
drastically). Englewood Cliffs: Educational Technology Publica-
• The best use of computers in education is tions, 1976.
as TOOLS- word processing, spreadsheets, Lewis, B.N. Decision Logic Tables for Algorithms and
Logicai Trees. London: HMSO, 1970.
and so on (in his view, this is not the best
Lewis, B.N., Horabin, I.S., and Gane, C.P. Flow Charts,
use, but is presented to teachers as such,
Logicai Trees and Algorithms for Rules and Regula-
since it is the least threatening use and is tions. London: HMSO, 1967.
thus a way of "getting a foot in the door"). Lewis, B.N., and Wolfenden, P.J. Algorithms and Logicai
If teachers as a body were to perceive the tools- Trees : A Self-Instructional Course. Cambridge, UK:
use of expert systems (and the tutee approach as Algorithms Press, 1969.
well) as "getting a foot in the door," so as to Lippert, R., and Trollip, S. Building Knowledge Bases for
eventually introduce some more role-threatening Expert Systems. Paper presented at the 28th Interna-
tional Conference of ADCIS, the Association for the
forms of "intelligent tutoring systems," then may-
Development of Computer-based Instructional Systems,
be progress will be much slower than here pre-
Washington, DC, November 1986.
dicted. But maybe they will perceive the potential
Mager, R.F., and Pipe, P. Analyzing Performance Problems.
educational benefits and see themselves in full
Belmont, CA: Fearon, 1970.
control as managers of the resources at theirMiller,dis- G.A. The Magic Number 7- Plus or Minus 2. In C.A.
posal. In that case, we may expect some interesting Miller, The Psychology of Communication. London:
innovations in the classroom. □ Penguin, 1967.
McAleese, R. The Representation of Knowledge in Au-
thoring Environments. In N. Rushby and A. Howe
References (Eds.), Aspects of Educational Technology- XIX:
Educational, Training, and Information Technologies-
Carbonell, J.R. Al in CAI: An Artificial Intelligence Ap- Economics and Other Realities. London: Ko gan Page,
proach to Computer Assisted Instruction. IEEE Trans- 1986.
actions on Man-Machine Systems , November 1970. Merrill, M.D. Where Is the Authoring in Authoring Systems?
Dean, P.M. CBI: Some Widely Held Assumptions. The Journal of Computer-Based Instruction , 1985, /5(2).
Dean Lecture, presented at the 28th International Con- Merrill, M.D., and Wood, L.E. Computer Guided Instruc-
ference of the Association for the Development of tional Design. Journal of Computer Based Instruction ,
Computer-based Instructional Systems (ADCIS), Wash- Spring 1984, //(2).
ington, DC, November 1986. Newquist, H.P. Expert Systems: The Promise of a Smart
Dear, B.L. Artificial Intelligence Techniques: Applications Machine. Computerworld , January 1986.
for Courseware Development. Educational Technology , Papert, S. M ind storms: Children, Computers, and Power-
July 1986, 26(7), 7-15. ful Ideas. New York: Basic Books, 1980.
Gilbert, T.F. Praxeonomy: A Systematic Approach to Park, O., and Seidel, R.J. ICAI: Intelligent Applications
Identifying Training Needs. Management of Personnel of Al. Paper presented at the 28th International Con-
Quarterly y 1967, 6(3). ference of ADCIS, the Association for the Development
Harless, J. An Ounce of Analysis (Is Worth a Pound of of Computer-based Instructional Systems, Washington,
Objectives). Newnan, GA: Harless Performance Guild, DC, November 1986.
1968. Pask, G. Review of Conversation Theory and a Protologic
Harmon, P. Expert Systems, Job Aids, and the Future (or protolanguage)- Lp. ERICJECTJ Annual Review
of Instructional Technology. Performance and Instruc- Paper. Educational Communications and Technology
tion Journal , March 1986. Journal ( ECTJ ), January 1984,52(1).
Harmon, P., and King, D. Expert Systems: Artificial In-Pipe, P. Decision Tables: The Poor Person's Answer to
telligence in Business . New York: John Wiley and Expert Systems. Performance and Instruction Journal ,
Sons, 1985. March 1986.
Horabin, I.S. Towards Greater Employee Productivity . Romiszowski, A.J. Selection and Use of Teaching Aids.
Summit Point, WV: 1971 . London: Kogan Page, 1968.
Horn, R.E. Course Notes for Information Mapping Work- Romiszowski, A.J. Classifications, Algorithms, and Check
Instructional Systems (ADCIS), Washington, DC, No- Lists as Aids to the Selection of Instructional Methods
vember, 1986a. and Media. In A. Bajpai and J. Leedham (Eds.) , Aspects
Kearsley, G. CBT Analyst- DISKETTE and Costs/Bene- of Educational Technology. Volume IV. London,
fits- DISKETTE. (Expert Systems). Lajolla, CA: Park Pitman, 1970.
Row Software, 1986b. Romiszowski, A.J. Selection and Use of Instructional
Instructional Systems (ADCIS), Washington, DC, No- Media: A Systems Approach. London: Kogan Page,
vember 1986a. 1974.
Kearsley, G. CBT Analyst- DISKETTE and Costs/Bene-Romiszowski, A.J. Problem Solving in Instructional Design:
fits- DISKETTE. (Expert Systems). Lajolla, CA: Park A Heuristic Approach. In A. Howe (Ed.), International
Row Software, 1986b. Yearbook of Educational and Instructional Technol-

EDUCATIONAL TECHNOLOGY/October 1 987 29

This content downloaded from

154.120.242.82 on Wed, 08 Jun 2Thu, 01 Jan 1976 12:34:56 UTC
All use subject to https://about.jstor.org/terms
 ogy- 1 980/81 ģ London: Kogan Page, 1980.
Romiszowski, A.J. Designing Instructional Systems : Deci-
sion-Making in Course Planning and Curriculum Design .
London: Kogan Page, 1981.
Romiszowski, A.J. Artificial Intelligence and Expert Sys-
tems in Education: Progress, Promise, and Problems.
Paper presented to EDTECH '86, the International
Conference of the Australian Society for Educational
Technology, Perth, Western Australia, December 1986.
Romiszowski, A.J. Artificial Intelligence and Expert
Systems in Education: Potential Promise or Threat to
Teachers? Educational Media International , 1987, in
press.
Singer, M.J., and Perez, R.S. A Demonstration of an Ex- Wilson, B.G., and Welsh, J.R. Small Knowledge-Based
pert System for Training Device Design. Journal of
Computer-Based Instruction . Spring 1986, /3(2).
Sleeman, D.H., and Brown, J.S. Intelligent Tutoring Sys-
tems. New York: Academic Press, 1982.
Starfield, A.M., Butala, K.L., England, M.M., and Smith,
K.A. Mastering Engineering Concepts by Building an
Expert System. Engineering Education} November
1983.
Taylor, R.P. (Ed.). The Computer in the School : Tutor ,
Too!. Tutee. New York: Teachers College Press, 1980.
Texas Instruments, Inc. Persona! Consultant : Expert
System Development Tools. User's Guide (manual)
and Software (for PC), 1985.
Wheatley, D.M., and Unwin, A.W. The Algorithm Writer's
Guide. London: Longman, 1972.
Systems in Education and Training: Something New
Under the Sun. Educational Technology , November
1986, 26(11), 7-1 3.

Events
• Florida Instructional Computing Con- • Computers in Education Conference.
ference. The Eighth Annual Florida The 1988 Computers in Education Con-
Instructional Computing Conference is ference sponsored by Arizona State Uni-

Calendar
• Data Training Conference and Expo.
The Eighth Annual Data Training
Conference and Expo sponsored by
Data Training Magazine wili be held
November 8-12, 1987 in Boston, Mas-
sachusetts. More than 60 sessions will
examine topics related to data process-
ing training. Contact: Julia Stasio,
Weingarten Publications, 38 Chauncey
Street, Boston, Massachusetts 02111;
(617) 542-0146.
• Pennsylvania Learning Resources Con-
ference. The annual meeting of the
Pennsylvania Learning Resources Asso- Coordinator, North Cook Educational
ciation, affiliated with both AECT and Service Center, 2701 Central Road,
ICCE, will be held November 23-24 in
Hershey, Pennsylvania. Topics will in- 5065.
clude special computer applications,
desktop publishing, public domain soft-
ware swaps, interactive video, and
optical disk technologies. Contact:
Angelo Speziale, Bethlehem Area
School District, IMC, 240 Elizabeth
Avenue, Bethlehem, Pennsylvania 18018;
865-6263.
• Optical Information Systems Conference.
The Annual Optical Information Sys:
terns Conference and Exposition will be • Technological Literacy Conference. The
held December 1-3, 1987 in New York
City. Sponsored by Meckler Corpora-
tion, the 1986 session, held in Washing-
ton, DC, attracted some 1900 regis-
trants and 93 exhibitors. This year's
themes focus on software, actual end-
user experiences, new developments,
and future needs. Contact: Marilyn
Reed, Meckler Corporation, 11 Ferry STS Program, Penn State University,
Lane West, Westport, Connecticut 06880; 128 Willard Building, University Park,
(800) 635-5537.
scheduled for January 25-28 in Orlando, versity's College of Education will be
Florida. The sponsors include the
Florida Department of Education and
held March 6-8,1988 in Tempe, Arizona.
The theme is: The Emerging Frontier:
several media and curriculum associa- Interactive Video, Artificial Intelligence,
tions. The conference agenda includes and Classroom Technology." Contact:
Gary Bitter, Arizona State University,
presentation sessions, exhibits, and work-
shops. Contact: Educational Technology Tempe, Arizona 85287; (602) 965-
7363.
Section, Florida Department of Educa-
tion, Knott Building, Tallahassee, Flori-
da 32399; (904) 488-0980. • Call for Presenters. A Call for presenters
has been issued for the Great Lakes
LOGO Conference to be held May 6,
1988. Seymour Papert will be the
• Computer Conference. MICRO/IDEAS
keynote speaker at the conference.
hosts its eighth annual computer con-
Contact: Educational Computer Con-
ference, "THE ROLE OF THE COM-
PUTER IN EDUCATION VIII," Febru- sortium of Ohio, 1123 S.O.M. Center
ary 24-26 1988 at the Woodfield Hilton Road, Cleveland, Ohio 44124. Presen-
and Towers, Arlington Heights, Illinois. tations can include research papers,
Contact: Richard F. Nelson, Conference reports of classroom experiences, or
ideas for the use of the LOGO com-
puter language.
Glenview, Illinois 60025; (312) 998-
• Bell Association Convention. The Bi-
ennial International Convention of the
Alexander Graham Bell Association for
• Teacher Education Meeting. The Asso- the Deaf will be held in Orlando, Flori-
ciation for Teacher Educators will hold da, July 19-23, 1988. The theme is
its annual National Forum for Teacher "Reach for the Stars," reflecting the
Education February 14-17, 1988 in San group's commitment to realizing the
Diego, California. Contact: Ann Karmos,potential of hearing-impaired children.
Planning Committee Chair, Southern Contact: Susan Coffman, Alexander
Illinois University, Carbondale, Illinois Graham Bell Association for the Deaf,
62901; (618) 536-2441. 3417 Volta Place NW, Washington,
DC; (202) 337-5220 (voice/TDD).
• Vocational Education Conference in
Third National Science, Technology, Australia. The TAFE National Centre
Society (STS) Conference, devoted to for Research and Development is to
Technological Literacy, will be held
sponsor an international conference
February 5-7, 1988 in Arlington, Vir- in Adelaide, Australia, March 12-19,
ginia. This year's theme will be "Tech- 1989. The theme of the meeting is
nology, Democracy, and Development." recent research in vocational educa-
Included will be the inaugural meeting tion. Contact: Denise Bubner, TAFE,
of the new Society for STS. Contact: 296 Payneham Road, Payneham, South
Australia 5070; (08) 42 7905.
Pennsylvania 16802. Your Listings Are Welcome

30 EDUCATIONAL TECHNOLOGY/October 1987

This content downloaded from

154.120.242.82 on Wed, 08 Jun 2022 08:29:51 UTC
All use subject to https://about.jstor.org/terms