Dr. Steve Knode
Information Resources Management College (IRMC)
Ft. McNair, Washington, D.C. 20319
202 685-2070
knode@ndu.edu

Professor Jon-David W. Knode
College of Notre Dame
Baltimore, MD 21210
410 675-6426
jdknode@hotmail.com

Abstract

Intelligent agents are rapidly becoming a way to leverage scarce academic resources to assist students. Agents today can search for information, inform and notify about key events, manage information, even provide information. Indeed, we are close to having real-time learner relationship management (LRM) agents suitable for providing a credible assistant faculty member.

Within the last few years the development and deployment of intelligent agents has become easier and cheaper. Although still an emerging technology, intelligent agents have already reached to point where they can be used to provide a means to enhance learning, whether in the classroom or via distance education. Several low-cost, easy-to-use packages have emerged, with useful features, such as anthropomorphic features and speech recognition. Soon, LRM agents will be providing useful information in many classrooms and over the web. This presentation will discuss several key developments in the intelligent agent CRM/LRM area. Several intelligent agents will assist in the presentation of material and will be demonstrated.

Introduction and Background

One of the most interesting, and potentially important, aspects of the information technology revolution might be the rapid integration of intelligent agent tutors into the classroom environment. "Intelligent software agents", also known as "knowbots" or just "bots" have recently become more than just a better way to find, filter and fuse information, especially from the internet. More and more agent technology is progressing to a point where the use of agents is becoming a main way in which to adjust to "information overload". By using intelligent agent technology, users are able to have multiple "assistants" available to provide them with the means to providing key information in a timely manner. Agent technology is the key to changing the metaphor of the internet from "self-service" to "room service". The difference in the metaphor is critical to dealing with the vast information base growing daily.

Intelligent agents (a.k.a. software agents and/or knowbots) are, in simplest terms, software programs built to autonomously perform certain specific tasks for the user. Intelligent Agents combine some elements of Artificial Intelligence (e.g., expert systems, object-oriented technology, neural networks, genetic algorithms, fuzzy logic, case-based reasoning, speech recognition, natural language processing, etc.) into an interface that has great applicability and appeal. Behind the agent front end is more often than not some simple rule-based programming which triggers actions when certain conditions are met. Newer agents are even beginning to evidence a crude capability to adjust their "reasoning" and "learn" from the user.

Today

Today, one growing and important use of intelligent agents is providing customer relationship management (CRM) or, in the case of educational institutions, learner relationship management (LRM). Already, agents are beginning to provide the same (better?) level of support for customers that has heretofore been provided by humans. Agents, for example, can:
Answer questions Often there is a relatively small set of questions which gets asked repeatedly. Anticipating these questions, agents can be scripted to provide accurate and timely answers. Instructors with experience teaching courses are easily able to provide the answers to these common questions and thereby reduce the burden of answering common questions.
Provide navigation assistance Especially in a distance education course, there are often questions about navigating a web page or finding the correct link to key information. Again, agents or "bots" can very readily provide this assistance.
Generate email responses Another useful function of an LRM agent is to generate an automatic email response. Today, many email packages support autoresponders, but these autoresponders are often simple "Out of the office messages". Soon, agents will be able to analyze email generated by students and respond appropriately with valuable answers, not simple administrative replies.
Advise on products and offerings Agents are quite capable of assisting in the selling aspects of products, be the products commercial or educational (courses or programs). By having a permanent sales agent available, visitors can get the information they need, whether it is a phone number or advice about applying to a program or information about a product.
Instruct and sympathize Agents can even assist in the instruction and support mode. With streaming video and audio, agents can provide instructional support almost at the level of an undergraduate teaching assistant in many aspects.Furthermore, agents are now beginning to provide students with just-in-time learning, customized for what they need at the moment they need it. These agent "instructors" can substitute for limited human instructors and, although not nearly as capable (yet!), can often answer a significant number of anticipated queries.

The advantages of using such a technology are numerous. Intelligent agents, for example, do not skip steps in helping customers, but show near infinite patience in answering the same or similar queries over and over. Agents do not (unless scripted to do so) have tempers or short fuses, so they can present a pleasant personna at all times. Further agents can be inexpensive to deploy, especially when compared to the cost of providing similar human assistance. Properly utilized, agents can be a formidable adjunct to a live education provider. Finally, an overlooked advantage is that of "dialogue mining". Agents record all conversations, so later these conversational logs can be examined and, if questions are being asked frequently which the agent has no answer for, these answers can be added to the knowledge base of the agent. In this manner, the agent "learns" or at least becomes smarter.

Future

The future is even brighter for the use of intelligent software tutors. Soon, however, it may be possible to have a personal tutor for each student. Running over the web, this intelligent agent tutor will be customized to not only provide the instruction that the student seeks at the time he or she seeks it, but will do so in a very effective manner. Several low cost, easy-to-use packages have begun to emerge, with useful features, such as impressive anthropomorphic features and speech recognition. Indeed, they could soon even become mentors for faculty as well as tutors for students. Already, inexpensive graphical interfaces exist which would allow a professor to put his or her own 3D face on the agent, thereby adding to the comfort level of the student.

As the use of the web increases and the concept of anytime, anywhere just-in-time learning becomes more popular, the use of intelligent agents to provide some instructional support will grow. As the technology rapidly matures, the limiting factor will no longer be the technology itself, but the development of a new paradigm for teaching. Instead of "one size fits all" lectures, for example, lectures will be customized for each student and provide information at a pace appropriate for each student. This mass customization feature has been taking place in other areas, but, so far, has been limited in education. Intelligent Agents will provide the means to have a personal tutor or instructor tailored for each student. Instead of having to rely on one instructor, students will have the ability to tap the information reservoir through their intelligent agents who can link to virtually all information. Instead of having to wait for feedback on an exam or private counseling session, students will instantly know what they have done right and wrong. Instead of having to meet at a particular time and place, students will be able to meet anywhere, anytime with their personal agent tutor and still enjoy the richness of a face-to-face session.

Additionally, there will be an increased development in the capability of the software agents to provide support. Soon, agents might be able to learn for themselves rather than being constantly reprogrammed with new information. Agents might, for example, be able to search the web for answers to questions that they cannot answer currently and, automatically, add these answers to their knowledge base. Increasing ability to "learn" from experiences, increased improvements in anthropomorphic features, increased friendliness, increased ability to collaborate and increased ease-of-use will no doubt boost the use of intelligent classroom agents.