An Intelligent Tutoring System (ITS) can exist in three very general types:
- An ITS can be a computer delivered service which enables and possibly controls and guides human to human tutoring activities so that the set of human teacher, human student, and computer-selected learning materials achieve better educational outcomes than were taking place human-to-human without computer assistance.
- An ITS may also select "educationally appropriate" pre-existing human-authored alternative learning material for a student of a particular subject domain and then enable and control the automated delivery to the student of that tutorial information stored on a computer. Ideally the delivery will be in ways most attractive or preferable to the student to maximize the student's learning of the subject matter.
- And in the "best case", an ITS may learn (a) how "similar students" learn and fail to learn, (b) learn from the human-authored learning materials, and then (c) dynamically generate the most appropriate tutorial instructional materials and tutorial experiences for an individual student on the demand of the student. That is, the ITS may generate these highly individualized materials "just in time" just when the student requests them or just when computer sensors determine the student needs the extra help of hints and other additional instruction about a topic which the student is studying. Few ITSes can yet perform the latter feat.
The "intelligent" part of an ITS is "artificially intelligent" software running on a "server computer". The AI software encapsulates a great deal, but not all, of the expert knowledge and reasoning abilities of human experts in one or more special subject matter fields of knowledge or subject domains. It also imitates the thinking or reasoning of the human experts who "taught" the ITS the typical facts, concepts and reasoning activities that exist in the subject matter areas about which the humans are expert. The special nature of AI software is that eventually it may come to "it's own conclusions" and take actions which the humans who taught it never thought of. And those conclusions and actions may be very appropriate and/or timely -- and therefore useful -- according to the humans who taught it.
Usually an ITS reflects the accumulated (or 'input') human expertise in a certain subject matter area, or field or domain of human knowledge, experience, behaviors, and sets of problems to solve that less expert students or practitioners of the subject matter want to learn to master. Thus the software expresses an "artificial intelligence" similar to that of a human experts who taught it "to reason" artificially and perhaps also respond "mechanically". Therefore, this component of an ITS is called the "AI" component.
(Note to self: add something about the "brittleness" of historic AI programs which "break" and can no longer "reason well" when the human user puts to the AI program a question or situation just "a little bit outside" the subject domain in which the AI program specializes.)
