Intelligent tutoring system

The educational technology and digital learning wiki
Jump to: navigation, search

See also: Artificial intelligence and education (AI&Ed), a set of subfields in educational technology intrested by artificial intelligence technology in education.

1 Definition

  • An Intelligent Tutoring System (ITS) replaces a human tutor by a machine. In most cases, it's "intelligent" CBT.
  • Research in ITS organizes the "problem" in (1) knowledge about a domain, (2) knowledge about the learner, (") and pedagogy (knowledge of teaching strategies). The major components of a typical ITS are therefore an expert (or domain) model, student model and tutoring model. The expert model should be able to solve the problems the tutoring module submits to the students. The tutor module controls the interaction with the student, based on its teaching knowledge and comparisons between the student model and the domain knowledge. The student model reflects what the machine can infer about the student's cognitive state.


2 History

Intelligent Tutoring Systems (ITS) came into existence in the mid-seventies and had their peak time in the late eighties at the same time when expert systems were in vogue.

During the nineties, most research in artificial intelligence and education, rather focused on intelligent learning environments more influenced by CBL, microworlds, etc. than CBT.

Although ITS research is still an active sub-field of research it went out of fashion, probably for several reasons. ITS systems were not cost effective enough to survive in training. In addition, building an intelligent system is really tough, e.g. student modelling and user interaction are research problems by themselves. However, the major reason why most researchers moved on (e.g. to CSCL) is best explained by a longer quote from Reeves 1998:

ITS attracted much more attention, funding, and research a few years ago than they do today. One telling sign is that the Journal of Artificial Intelligence in Education recently changed its name to the Journal of Interactive Learning Research. Even those who have been most involved in research and development targeted at producing "intelligent tutors" have begun to acknowledge the lack of impact they have had on mainstream education (Lajoie & Derry, 1993). A major factor contributing to the lack of success of ITS is that the technical difficulties inherent in building student models and facilitating human-like communications have been greatly underestimated by proponents of this approach.
In the face of the disappointing results of ITS, some experts suggest that "...the appropriate role for a computer is not that of a teacher/expert, but rather, that of a mind-extension 'cognitive tool'" (Derry & Lajoie, 1993, p. 5). Cognitive tools [...] are unintelligent tools, relying on the learner to provide the intelligence, not the computer. This means that planning, decision-making, and self-regulation are the responsibility of the learner, not the technology. Cognitive tools can serve as powerful catalysts for facilitating these higher order skills if they are used in ways that promote reflection, discussion, and collaborative problem-solving [...].

In the late 2010s, ITS gained again interest, in particular since "statistical AI" based on neural networks and related techniques allows extracting patterns from big data. These can be for example to recommender systems that give advice based on the behavior of many or learning analytic systems that detect students with difficulties. With supervised learning (e.g. experts train a system by tagging examples) such systems also can learn to distinguish between good and less student productions.

3 Future (beyond 2010)

  • Integration in social learning environments (e.g. through various kinds of agents)
  • Things related to the semantic web (whenever that will happen ...)
  • Emotional computing
  • Intelligent devices (e.g. table tops)
  • Distributed cognition (over people and devices)
  • AI-enhanced e-learning systems (e.g. adaptive systems that select learning paths)

4 Software

There is not much ready-to-use software available. Several ITS systems were built on top of expert system technology

Below are a few links to commercial adaptive learning technology that can be use to build adaptive environments:

5 Links

6 References

  • Derry, S. J., & Lajoie, S. P. (1993). A middle camp for (un)intelligent instructional computing: An introduction. In S. P. Lajoie & S. J. Derry (Eds.), Computers as cognitive tools (pp. 1-11). Hillsdale, NJ: Lawrence Erlbaum.
  • Lajoie, S. P., & Derry, S. J. (Eds.). (1993). Computers as cognitive tools. Hillsdale, NJ: Lawrence Erlbaum.
  • Reeves, Thomas C. (1998), The Impact of Media and Technology in Schools, A Research Report prepared for The Bertelsmann Foundation, PDF

Here is a list of readers, books or articles that provide introductions:

  • Costa, E. (Ed.) (1992). New Directions for Intelligent Tutoring Systems (pp 15-27). Berlin: Springer-Verlag.
  • Goodyear, P. (Ed) (1991). Teaching knowledge and intelligent tutoring. Norwood (NJ): Ablex.
  • Mandl and A. Lesgold (Eds) (1988), Learning Issues for Intelligent Tutoring Systems. New York: Springer Verlag.
  • Self, J.A. (1986) The application of machine learning to student modelling. Instructional Science, 14, 327-338.
  • Sleeman D.H. and J.S. Brown (Eds.) (1998) Intelligent Tutoring Systems. Academic Press, London.
  • Wenger, Etienne (1987). Artificial Intelligence and Tutoring Systems. Morgan Kaufmann, Los Altos, CA, 1987. This probably remains the best book in the area, although it contributed to a paradigm shift from ITS towards cognitive tools.